Engines
64 engines, any two of them as a pair, and what each can do.
Every engine this version has, and what it can do - read off the code, so this page claims no more and no less than it does. A hop names its two sides by their addresses; migkit's own decision engine finds the fastest, exact, consistent way for each source -> target pair, and says which it chose and why.
The engines
| Engine | Name | Also written | Address schemes | Role |
|---|---|---|---|---|
| PostgreSQL | postgres | alloydb, aurora-postgres, pgsql, postgresql | postgres://, aurora-postgres://, alloydb://, postgresql://, pgsql:// | either side |
| MySQL | mysql | aurora-mysql, mariadb, percona, tdsql, tdsql-c | mysql://, mariadb://, percona://, aurora-mysql://, tdsql://, tdsql-c:// | either side |
| MongoDB | mongodb | cosmosdb-mongo, documentdb | mongodb://, documentdb://, cosmosdb-mongo://, mongodb+srv:// | either side |
| SQL Server | mssql | azure-sql, sqlserver | mssql://, azure-sql://, sqlserver:// | either side |
| Redis | redis | valkey | redis://, valkey://, rediss:// | either side |
| Kafka | kafka | - | kafka:// | either side |
| SQLite | sqlite | - | sqlite:// | either side |
| Parquet files | parquet | - | parquet://, parquet+s3:// | either side |
| ClickHouse | clickhouse | - | clickhouse:// | either side |
| DynamoDB | dynamodb | - | dynamodb:// | either side |
| Oracle Database | oracle | - | oracle:// | either side |
| Db2 (Linux, UNIX, Windows) | db2 | - | db2:// | either side |
| Db2 for i | db2i | as400, db2-for-i, ibmi, iseries | db2i://, ibmi://, as400://, iseries://, db2-for-i:// | either side |
| Db2 for z/OS | db2zos | db2-zos, db2z | db2zos://, db2-zos://, db2z:// | either side |
| SAP ASE | ase | sap-ase, sybase | ase://, sybase://, sap-ase:// | either side |
| Teradata | teradata | td, vantage | teradata://, td://, vantage:// | either side |
| OpenSearch | opensearch | elasticsearch | opensearch://, elasticsearch:// | either side |
| Cassandra | cassandra | scylla, scylladb | cassandra://, scylladb://, scylla:// | either side |
| Amazon Redshift | redshift | - | redshift:// | either side |
| Snowflake | snowflake | - | snowflake:// | either side |
| BigQuery | bigquery | - | bigquery:// | either side |
| Kinesis Data Streams | kinesis | - | kinesis:// | a stream a pair delivers its changes into |
| Google Pub/Sub | pubsub | - | pubsub:// | a stream a pair delivers its changes into |
| DuckDB | duckdb | - | duckdb:// | either side |
| Files described by COBOL copybooks | copybook | cobol, vsam | copybook://, vsam://, cobol:// | either side |
| Iceberg tables | iceberg | - | iceberg://, iceberg+http://, iceberg+https:// | either side |
| Delta tables | delta | delta-lake | delta://, delta-lake://, delta+s3:// | either side |
| TiDB | tidb | - | tidb:// | either side |
| CockroachDB | cockroachdb | cockroach, crdb | cockroachdb://, cockroach://, crdb:// | either side |
| Aurora DSQL | dsql | aurora-dsql | dsql://, aurora-dsql:// | either side |
| Couchbase | couchbase | - | couchbase://, couchbases:// | either side |
| Cosmos DB for NoSQL | cosmosdb | azure-cosmos, cosmos, cosmosdb-nosql | cosmosdb://, cosmos://, cosmosdb-nosql://, azure-cosmos:// | either side |
| Google Cloud Spanner | spanner | cloud-spanner, spanner-pg | spanner://, cloud-spanner://, spanner-pg:// | either side |
| Google Cloud Bigtable | bigtable | cloud-bigtable | bigtable://, cloud-bigtable:// | either side |
| Firestore (native mode) | firestore | firestore-native | firestore://, firestore-native:// | either side |
| Firestore in Datastore mode | datastore | cloud-datastore, firestore-datastore | datastore://, firestore-datastore://, cloud-datastore:// | either side |
| GaussDB (openGauss) | gaussdb | gaussdb-opengauss, opengauss | gaussdb://, opengauss://, gaussdb-opengauss:// | either side |
| OceanBase (MySQL mode) | oceanbase | oceanbase-mysql | oceanbase://, oceanbase-mysql:// | either side |
| OceanBase (Oracle mode) | oceanbase-oracle | - | oceanbase-oracle:// | either side |
| PolarDB-X | polardbx | drds, polardb-x | polardbx://, polardb-x://, drds:// | either side |
| TDSQL (distributed) | dcdb | tdsql-distributed, tdsql-shard | dcdb://, tdsql-distributed://, tdsql-shard:// | either side |
| Amazon Keyspaces | keyspaces | amazon-keyspaces, aws-keyspaces | keyspaces://, amazon-keyspaces://, aws-keyspaces:// | either side |
| Amazon Neptune | neptune | amazon-neptune | neptune://, amazon-neptune:// | either side |
| Amazon Timestream | timestream | amazon-timestream | timestream://, amazon-timestream:// | either side |
| Lindorm | lindorm | - | lindorm:// | either side |
| GeminiDB (Cassandra API) | geminidb-cassandra | geminidb | geminidb-cassandra://, geminidb:// | either side |
| GeminiDB (Redis API) | geminidb-redis | - | geminidb-redis:// | either side |
| GeminiDB (MongoDB API) | geminidb-mongo | - | geminidb-mongo:// | either side |
| Qdrant | qdrant | - | qdrant:// | either side |
| Neo4j | neo4j | neo4j-aura | neo4j://, neo4j-aura:// | either side |
| AnalyticDB for MySQL | analyticdb-mysql | adb-mysql | analyticdb-mysql://, adb-mysql:// | either side |
| AnalyticDB for PostgreSQL | analyticdb-postgres | adb-pg, analyticdb-pg | analyticdb-postgres://, adb-pg://, analyticdb-pg:// | either side |
| Azure Synapse (dedicated SQL pool) | synapse | azure-synapse | synapse://, azure-synapse:// | either side |
| YugabyteDB | yugabytedb | ysql, yugabyte | yugabytedb://, yugabyte://, ysql:// | either side |
| SingleStore | singlestore | memsql | singlestore://, memsql:// | either side |
| Netezza | netezza | pure-data-netezza | netezza://, pure-data-netezza:// | either side |
| SAP HANA | saphana | hana, sap-hana | saphana://, hana://, sap-hana:// | either side |
| Informix | informix | - | informix:// | either side |
| Progress OpenEdge | progress | openedge, progress-openedge | progress://, openedge://, progress-openedge:// | either side |
| Vertica | vertica | - | vertica:// | either side |
| Exasol | exasol | - | exasol:// | either side |
| Databricks SQL | databricks | databricks-sql | databricks://, databricks-sql:// | either side |
| SAP application data (ODP) | sapodp | odp, sap-odp | sapodp://, sap-odp://, odp://, sapodp+http:// | a source, read and never written |
| SaaS applications, read by a connector program | saas | program | - | a source, read and never written |
| Any pair of two engines | hetero | - | - | the pair: a source of one engine, a target of another |
| Any other SQL engine | generic | - | trino://, presto:// | either side, compared by the tables the hop lists |
What each can do
yes: the code does it for this engine. n/a: it does not apply, and the reason is said. not yet: not done yet, and what to do meanwhile is said. Hover a cell for its reason.
schema: comparing the schemacounts: comparing row countsdata: comparing the data itselfdeep: the deep checkssequences: carrying sequences and auto-increment valuesparams: comparing server settingsbulk-move: moving a whole database in bulktable-copy: copying table by table, resumablystream: keeping the target following the sourcefence: proving the target has caught up before cutoverconfirm: telling a difference still arriving from one that is wrongdelta: verifying only what changedusers: carrying users and their grantsguard: noticing a move that moved nothingstatistics: refreshing the target's statistics after a loadsnapshot: snapshotting the target so a cutover can be rolled back
| Engine | schema | counts | data | deep | sequences | params | bulk-move | table-copy | stream | fence | confirm | delta | users | guard | statistics | snapshot |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PostgreSQL | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes |
| MySQL | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes |
| MongoDB | yes | yes | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | yes | yes | n/a | yes |
| SQL Server | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes |
| Redis | yes | yes | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | yes | yes | n/a | yes |
| Kafka | yes | yes | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | yes | yes | n/a | yes |
| SQLite | yes | yes | yes | yes | yes | yes | yes | yes | n/a | n/a | n/a | n/a | n/a | yes | yes | yes |
| Parquet files | yes | yes | yes | yes | n/a | n/a | yes | yes | n/a | n/a | n/a | n/a | n/a | yes | n/a | yes |
| ClickHouse | yes | yes | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | yes | yes | n/a | yes |
| DynamoDB | yes | yes | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | n/a | yes | n/a | yes |
| Oracle Database | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes |
| Db2 (Linux, UNIX, Windows) | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes |
| Db2 for i | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | n/a | yes |
| Db2 for z/OS | yes | yes | yes | yes | yes | yes | yes | yes | n/a | n/a | n/a | n/a | yes | yes | yes | yes |
| SAP ASE | yes | yes | yes | yes | yes | yes | yes | yes | not yet | yes | yes | yes | yes | yes | yes | yes |
| Teradata | yes | yes | yes | yes | n/a | n/a | n/a | yes | n/a | n/a | n/a | n/a | yes | yes | yes | n/a |
| OpenSearch | yes | yes | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | yes | yes | n/a | yes |
| Cassandra | yes | yes | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | yes | yes | n/a | yes |
| Amazon Redshift | yes | yes | yes | yes | yes | yes | n/a | yes | n/a | n/a | n/a | n/a | yes | yes | n/a | n/a |
| Snowflake | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | n/a | yes |
| BigQuery | yes | yes | yes | yes | n/a | n/a | yes | yes | yes | yes | yes | yes | yes | yes | n/a | yes |
| Kinesis Data Streams | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a |
| Google Pub/Sub | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a |
| DuckDB | yes | yes | yes | yes | yes | n/a | yes | yes | n/a | n/a | n/a | n/a | n/a | yes | n/a | yes |
| Files described by COBOL copybooks | yes | yes | yes | yes | n/a | n/a | yes | yes | n/a | n/a | n/a | n/a | n/a | yes | n/a | yes |
| Iceberg tables | yes | yes | yes | not yet | n/a | n/a | not yet | yes | not yet | not yet | not yet | not yet | n/a | yes | n/a | not yet |
| Delta tables | yes | yes | yes | not yet | n/a | n/a | not yet | yes | not yet | not yet | not yet | not yet | n/a | yes | n/a | not yet |
| TiDB | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes |
| CockroachDB | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | n/a | yes |
| Aurora DSQL | yes | yes | yes | yes | yes | n/a | n/a | yes | yes | yes | yes | yes | n/a | yes | yes | yes |
| Couchbase | yes | yes | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes |
| Cosmos DB for NoSQL | yes | yes | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | n/a | yes | n/a | yes |
| Google Cloud Spanner | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes |
| Google Cloud Bigtable | yes | yes | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | n/a | yes | n/a | yes |
| Firestore (native mode) | yes | yes | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | n/a | yes | n/a | yes |
| Firestore in Datastore mode | yes | yes | yes | yes | n/a | yes | yes | yes | n/a | n/a | n/a | n/a | n/a | yes | n/a | yes |
| GaussDB (openGauss) | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes |
| OceanBase (MySQL mode) | yes | yes | yes | yes | yes | yes | not yet | yes | yes | yes | yes | yes | yes | yes | yes | yes |
| OceanBase (Oracle mode) | yes | yes | yes | yes | yes | yes | not yet | yes | not yet | not yet | not yet | not yet | yes | yes | yes | yes |
| PolarDB-X | yes | yes | yes | yes | yes | yes | not yet | yes | yes | yes | yes | yes | yes | yes | yes | yes |
| TDSQL (distributed) | yes | yes | yes | yes | yes | yes | not yet | yes | yes | yes | yes | yes | yes | yes | yes | yes |
| Amazon Keyspaces | yes | yes | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | n/a | yes | n/a | yes |
| Amazon Neptune | yes | yes | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes |
| Amazon Timestream | yes | yes | yes | n/a | n/a | yes | yes | yes | n/a | n/a | n/a | n/a | n/a | yes | n/a | yes |
| Lindorm | yes | yes | yes | yes | n/a | yes | n/a | yes | not yet | not yet | not yet | not yet | yes | yes | n/a | yes |
| GeminiDB (Cassandra API) | yes | yes | yes | yes | n/a | yes | yes | yes | n/a | n/a | n/a | n/a | yes | yes | n/a | yes |
| GeminiDB (Redis API) | yes | yes | yes | yes | n/a | yes | n/a | yes | yes | n/a | n/a | yes | yes | yes | n/a | yes |
| GeminiDB (MongoDB API) | yes | yes | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | yes | yes | n/a | yes |
| Qdrant | yes | yes | yes | yes | n/a | yes | not yet | yes | n/a | n/a | n/a | n/a | n/a | yes | n/a | not yet |
| Neo4j | yes | yes | yes | yes | n/a | yes | not yet | yes | yes | yes | not yet | yes | yes | yes | yes | not yet |
| AnalyticDB for MySQL | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | yes | yes |
| AnalyticDB for PostgreSQL | yes | yes | yes | yes | yes | not yet | n/a | yes | yes | yes | yes | yes | yes | yes | not yet | not yet |
| Azure Synapse (dedicated SQL pool) | yes | yes | yes | yes | n/a | yes | n/a | yes | n/a | n/a | n/a | n/a | yes | yes | yes | yes |
| YugabyteDB | yes | yes | yes | yes | yes | not yet | n/a | yes | yes | yes | yes | yes | yes | yes | not yet | not yet |
| SingleStore | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | yes | yes |
| Netezza | yes | yes | yes | yes | n/a | n/a | n/a | yes | yes | yes | yes | yes | n/a | yes | n/a | n/a |
| SAP HANA | yes | yes | yes | yes | n/a | n/a | n/a | yes | yes | yes | yes | yes | n/a | yes | n/a | n/a |
| Informix | yes | yes | yes | yes | n/a | n/a | n/a | yes | yes | yes | yes | yes | n/a | yes | n/a | n/a |
| Progress OpenEdge | yes | yes | yes | yes | n/a | n/a | n/a | yes | yes | yes | yes | yes | n/a | yes | n/a | n/a |
| Vertica | yes | yes | yes | yes | n/a | n/a | n/a | yes | yes | yes | yes | yes | n/a | yes | n/a | n/a |
| Exasol | yes | yes | yes | yes | n/a | n/a | n/a | yes | yes | yes | yes | yes | n/a | yes | n/a | n/a |
| Databricks SQL | yes | yes | yes | yes | n/a | n/a | n/a | yes | yes | yes | yes | yes | n/a | yes | n/a | n/a |
| SAP application data (ODP) | yes | yes | yes | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a |
| SaaS applications, read by a connector program | yes | yes | yes | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a |
| Any pair of two engines | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes |
| Any other SQL engine | yes | yes | yes | yes | n/a | n/a | n/a | yes | yes | yes | yes | yes | n/a | yes | n/a | n/a |
Each shape of table
A table without a key, a key of many columns, a table too large for one range: each shape has a way of its own on each engine, chosen by the decision engine.
keyless: a table without a keyno unique index: a table with no unique indexwide key: a key of many columns or of texttoo large: a table too large for one rangelarge values: a table with large valuespartitioned: a partitioned tableread-only source: a source that is read-only or a standbymissing grant: a user without the grant a program needsread-only target: a target that must stay read-onlyslow link: a link too slow for one streamwrites held: a side whose writes are held and let go again, read back - the source at cutoverspread target: a target that spreads a table across nodes of its own, its rows placed and loaded as it places them
| Engine | keyless | no unique index | wide key | too large | large values | partitioned | read-only source | missing grant | read-only target | slow link | writes held | spread target |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| PostgreSQL | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes |
| MySQL | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes |
| MongoDB | n/a | n/a | yes | yes | yes | n/a | yes | not yet | yes | yes | yes | yes |
| SQL Server | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes | n/a |
| Redis | n/a | n/a | n/a | yes | yes | n/a | yes | n/a | yes | yes | yes | n/a |
| Kafka | n/a | n/a | n/a | yes | yes | yes | yes | n/a | yes | yes | yes | n/a |
| SQLite | yes | yes | yes | yes | yes | n/a | yes | n/a | n/a | n/a | n/a | n/a |
| Parquet files | yes | yes | n/a | yes | yes | yes | yes | n/a | n/a | n/a | n/a | n/a |
| ClickHouse | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes | n/a |
| DynamoDB | yes | yes | n/a | yes | n/a | n/a | yes | n/a | yes | yes | n/a | n/a |
| Oracle Database | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | yes | n/a |
| Db2 (Linux, UNIX, Windows) | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes | n/a |
| Db2 for i | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | n/a | n/a |
| Db2 for z/OS | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | n/a | n/a |
| SAP ASE | yes | yes | yes | yes | yes | not yet | yes | n/a | yes | yes | yes | n/a |
| Teradata | yes | yes | n/a | yes | yes | yes | yes | yes | yes | yes | yes | n/a |
| OpenSearch | yes | yes | n/a | yes | yes | n/a | yes | n/a | yes | yes | yes | n/a |
| Cassandra | yes | yes | n/a | yes | n/a | n/a | yes | n/a | yes | yes | yes | yes |
| Amazon Redshift | yes | yes | yes | yes | n/a | n/a | yes | n/a | yes | yes | yes | n/a |
| Snowflake | yes | yes | yes | yes | yes | n/a | yes | n/a | yes | yes | yes | n/a |
| BigQuery | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | n/a | n/a |
| Kinesis Data Streams | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a |
| Google Pub/Sub | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a | n/a |
| DuckDB | yes | yes | yes | yes | yes | n/a | yes | n/a | n/a | n/a | n/a | n/a |
| Files described by COBOL copybooks | yes | yes | n/a | yes | n/a | n/a | yes | n/a | n/a | n/a | n/a | n/a |
| Iceberg tables | yes | yes | n/a | yes | yes | not yet | yes | n/a | n/a | yes | n/a | n/a |
| Delta tables | yes | yes | n/a | yes | yes | not yet | yes | n/a | n/a | yes | n/a | n/a |
| TiDB | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes | n/a |
| CockroachDB | yes | yes | yes | yes | yes | n/a | yes | n/a | yes | yes | yes | n/a |
| Aurora DSQL | yes | yes | yes | yes | yes | n/a | yes | n/a | yes | yes | n/a | n/a |
| Couchbase | n/a | n/a | n/a | yes | yes | n/a | yes | n/a | yes | yes | yes | n/a |
| Cosmos DB for NoSQL | n/a | n/a | n/a | yes | n/a | n/a | yes | n/a | n/a | yes | n/a | n/a |
| Google Cloud Spanner | n/a | n/a | yes | yes | n/a | n/a | yes | n/a | n/a | yes | n/a | n/a |
| Google Cloud Bigtable | n/a | n/a | n/a | yes | not yet | n/a | yes | n/a | n/a | yes | n/a | n/a |
| Firestore (native mode) | n/a | n/a | n/a | yes | n/a | n/a | yes | n/a | n/a | yes | n/a | n/a |
| Firestore in Datastore mode | n/a | n/a | n/a | yes | n/a | n/a | yes | n/a | n/a | yes | n/a | n/a |
| GaussDB (openGauss) | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes | n/a |
| OceanBase (MySQL mode) | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes | n/a |
| OceanBase (Oracle mode) | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes | n/a |
| PolarDB-X | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes | n/a |
| TDSQL (distributed) | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes | n/a |
| Amazon Keyspaces | yes | yes | n/a | yes | n/a | n/a | yes | n/a | n/a | yes | n/a | n/a |
| Amazon Neptune | n/a | n/a | n/a | yes | not yet | n/a | yes | n/a | n/a | yes | n/a | n/a |
| Amazon Timestream | n/a | n/a | n/a | yes | not yet | n/a | yes | n/a | n/a | yes | n/a | n/a |
| Lindorm | n/a | n/a | yes | yes | not yet | not yet | yes | n/a | yes | yes | n/a | n/a |
| GeminiDB (Cassandra API) | yes | yes | n/a | yes | n/a | n/a | yes | n/a | yes | yes | yes | n/a |
| GeminiDB (Redis API) | n/a | n/a | n/a | yes | yes | n/a | yes | n/a | yes | yes | yes | n/a |
| GeminiDB (MongoDB API) | n/a | n/a | yes | yes | yes | n/a | yes | n/a | yes | yes | yes | n/a |
| Qdrant | n/a | n/a | n/a | yes | not yet | not yet | yes | n/a | n/a | yes | n/a | n/a |
| Neo4j | n/a | n/a | n/a | yes | not yet | not yet | yes | n/a | n/a | yes | n/a | n/a |
| AnalyticDB for MySQL | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes | n/a |
| AnalyticDB for PostgreSQL | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes | n/a |
| Azure Synapse (dedicated SQL pool) | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes | n/a |
| YugabyteDB | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes | n/a |
| SingleStore | yes | yes | yes | yes | yes | yes | yes | n/a | yes | yes | yes | n/a |
| Netezza | not yet | not yet | not yet | yes | not yet | not yet | yes | n/a | yes | yes | yes | n/a |
| SAP HANA | not yet | not yet | not yet | yes | not yet | not yet | yes | n/a | yes | yes | yes | n/a |
| Informix | not yet | not yet | not yet | yes | not yet | not yet | yes | n/a | yes | yes | yes | n/a |
| Progress OpenEdge | not yet | not yet | not yet | yes | not yet | not yet | yes | n/a | yes | yes | yes | n/a |
| Vertica | not yet | not yet | not yet | yes | not yet | not yet | yes | n/a | yes | yes | yes | n/a |
| Exasol | not yet | not yet | not yet | yes | not yet | not yet | yes | n/a | yes | yes | yes | n/a |
| Databricks SQL | not yet | not yet | not yet | yes | not yet | not yet | yes | n/a | yes | yes | yes | n/a |
| SAP application data (ODP) | yes | yes | n/a | yes | n/a | n/a | yes | n/a | n/a | yes | n/a | n/a |
| SaaS applications, read by a connector program | yes | yes | n/a | yes | n/a | n/a | yes | n/a | n/a | yes | n/a | n/a |
| Any pair of two engines | yes | yes | yes | yes | yes | not yet | yes | not yet | yes | yes | yes | yes |
| Any other SQL engine | yes | yes | yes | yes | yes | yes | yes | n/a | n/a | yes | n/a | n/a |
Stored code carried and proved
Views, functions, procedures and triggers are converted one at a time by whichever way proves itself on each, and each version, whoever wrote it, is run on both sides in a sandbox and compared. Between any two of PostgreSQL, MySQL, SQL Server, SQLite, ClickHouse, Oracle Database, Teradata, BigQuery, TiDB, CockroachDB, Aurora DSQL, GaussDB (openGauss), OceanBase (MySQL mode), OceanBase (Oracle mode), PolarDB-X, TDSQL (distributed), AnalyticDB for MySQL, AnalyticDB for PostgreSQL, YugabyteDB, SingleStore, a routine is carried by:
- the operator's own file: view, function, procedure, trigger, event
- the query translator: view, function
- a model: view, function, procedure, trigger, event
and these pairs further, by migkit's own converters:
| Source | Target | What is converted |
|---|---|---|
| PostgreSQL | MySQL | view, function, procedure, trigger |
| PostgreSQL | TiDB | view, function, procedure, trigger |
| PostgreSQL | OceanBase (MySQL mode) | view, function, procedure, trigger |
| PostgreSQL | PolarDB-X | view, function, procedure, trigger |
| PostgreSQL | TDSQL (distributed) | view, function, procedure, trigger |
| PostgreSQL | AnalyticDB for MySQL | view, function, procedure, trigger |
| PostgreSQL | SingleStore | view, function, procedure, trigger |
| MySQL | PostgreSQL | view, function, procedure, trigger, event |
| MySQL | CockroachDB | view, function, procedure, trigger, event |
| MySQL | Aurora DSQL | view, function, procedure, trigger, event |
| MySQL | GaussDB (openGauss) | view, function, procedure, trigger, event |
| MySQL | AnalyticDB for PostgreSQL | view, function, procedure, trigger, event |
| MySQL | YugabyteDB | view, function, procedure, trigger, event |
| SQL Server | PostgreSQL | view, function, procedure, trigger, event |
| SQL Server | MySQL | view, function, procedure, trigger |
| SQL Server | TiDB | view, function, procedure, trigger |
| SQL Server | CockroachDB | view, function, procedure, trigger, event |
| SQL Server | Aurora DSQL | view, function, procedure, trigger, event |
| SQL Server | GaussDB (openGauss) | view, function, procedure, trigger, event |
| SQL Server | OceanBase (MySQL mode) | view, function, procedure, trigger |
| SQL Server | PolarDB-X | view, function, procedure, trigger |
| SQL Server | TDSQL (distributed) | view, function, procedure, trigger |
| SQL Server | AnalyticDB for MySQL | view, function, procedure, trigger |
| SQL Server | AnalyticDB for PostgreSQL | view, function, procedure, trigger, event |
| SQL Server | YugabyteDB | view, function, procedure, trigger, event |
| SQL Server | SingleStore | view, function, procedure, trigger |
| Oracle Database | PostgreSQL | view, function, procedure, trigger |
| Oracle Database | MySQL | view, function, procedure, trigger |
| Oracle Database | TiDB | view, function, procedure, trigger |
| Oracle Database | CockroachDB | view, function, procedure, trigger |
| Oracle Database | Aurora DSQL | view, function, procedure, trigger |
| Oracle Database | GaussDB (openGauss) | view, function, procedure, trigger |
| Oracle Database | OceanBase (MySQL mode) | view, function, procedure, trigger |
| Oracle Database | PolarDB-X | view, function, procedure, trigger |
| Oracle Database | TDSQL (distributed) | view, function, procedure, trigger |
| Oracle Database | AnalyticDB for MySQL | view, function, procedure, trigger |
| Oracle Database | AnalyticDB for PostgreSQL | view, function, procedure, trigger |
| Oracle Database | YugabyteDB | view, function, procedure, trigger |
| Oracle Database | SingleStore | view, function, procedure, trigger |
| TiDB | PostgreSQL | view, function, procedure, trigger, event |
| TiDB | CockroachDB | view, function, procedure, trigger, event |
| TiDB | Aurora DSQL | view, function, procedure, trigger, event |
| TiDB | GaussDB (openGauss) | view, function, procedure, trigger, event |
| TiDB | AnalyticDB for PostgreSQL | view, function, procedure, trigger, event |
| TiDB | YugabyteDB | view, function, procedure, trigger, event |
| CockroachDB | MySQL | view, function, procedure, trigger |
| CockroachDB | TiDB | view, function, procedure, trigger |
| CockroachDB | OceanBase (MySQL mode) | view, function, procedure, trigger |
| CockroachDB | PolarDB-X | view, function, procedure, trigger |
| CockroachDB | TDSQL (distributed) | view, function, procedure, trigger |
| CockroachDB | AnalyticDB for MySQL | view, function, procedure, trigger |
| CockroachDB | SingleStore | view, function, procedure, trigger |
| Aurora DSQL | MySQL | view, function, procedure, trigger |
| Aurora DSQL | TiDB | view, function, procedure, trigger |
| Aurora DSQL | OceanBase (MySQL mode) | view, function, procedure, trigger |
| Aurora DSQL | PolarDB-X | view, function, procedure, trigger |
| Aurora DSQL | TDSQL (distributed) | view, function, procedure, trigger |
| Aurora DSQL | AnalyticDB for MySQL | view, function, procedure, trigger |
| Aurora DSQL | SingleStore | view, function, procedure, trigger |
| GaussDB (openGauss) | MySQL | view, function, procedure, trigger |
| GaussDB (openGauss) | TiDB | view, function, procedure, trigger |
| GaussDB (openGauss) | OceanBase (MySQL mode) | view, function, procedure, trigger |
| GaussDB (openGauss) | PolarDB-X | view, function, procedure, trigger |
| GaussDB (openGauss) | TDSQL (distributed) | view, function, procedure, trigger |
| GaussDB (openGauss) | AnalyticDB for MySQL | view, function, procedure, trigger |
| GaussDB (openGauss) | SingleStore | view, function, procedure, trigger |
| OceanBase (MySQL mode) | PostgreSQL | view, function, procedure, trigger, event |
| OceanBase (MySQL mode) | CockroachDB | view, function, procedure, trigger, event |
| OceanBase (MySQL mode) | Aurora DSQL | view, function, procedure, trigger, event |
| OceanBase (MySQL mode) | GaussDB (openGauss) | view, function, procedure, trigger, event |
| OceanBase (MySQL mode) | AnalyticDB for PostgreSQL | view, function, procedure, trigger, event |
| OceanBase (MySQL mode) | YugabyteDB | view, function, procedure, trigger, event |
| OceanBase (Oracle mode) | PostgreSQL | view, function, procedure, trigger |
| OceanBase (Oracle mode) | MySQL | view, function, procedure, trigger |
| OceanBase (Oracle mode) | TiDB | view, function, procedure, trigger |
| OceanBase (Oracle mode) | CockroachDB | view, function, procedure, trigger |
| OceanBase (Oracle mode) | Aurora DSQL | view, function, procedure, trigger |
| OceanBase (Oracle mode) | GaussDB (openGauss) | view, function, procedure, trigger |
| OceanBase (Oracle mode) | OceanBase (MySQL mode) | view, function, procedure, trigger |
| OceanBase (Oracle mode) | PolarDB-X | view, function, procedure, trigger |
| OceanBase (Oracle mode) | TDSQL (distributed) | view, function, procedure, trigger |
| OceanBase (Oracle mode) | AnalyticDB for MySQL | view, function, procedure, trigger |
| OceanBase (Oracle mode) | AnalyticDB for PostgreSQL | view, function, procedure, trigger |
| OceanBase (Oracle mode) | YugabyteDB | view, function, procedure, trigger |
| OceanBase (Oracle mode) | SingleStore | view, function, procedure, trigger |
| PolarDB-X | PostgreSQL | view, function, procedure, trigger, event |
| PolarDB-X | CockroachDB | view, function, procedure, trigger, event |
| PolarDB-X | Aurora DSQL | view, function, procedure, trigger, event |
| PolarDB-X | GaussDB (openGauss) | view, function, procedure, trigger, event |
| PolarDB-X | AnalyticDB for PostgreSQL | view, function, procedure, trigger, event |
| PolarDB-X | YugabyteDB | view, function, procedure, trigger, event |
| TDSQL (distributed) | PostgreSQL | view, function, procedure, trigger, event |
| TDSQL (distributed) | CockroachDB | view, function, procedure, trigger, event |
| TDSQL (distributed) | Aurora DSQL | view, function, procedure, trigger, event |
| TDSQL (distributed) | GaussDB (openGauss) | view, function, procedure, trigger, event |
| TDSQL (distributed) | AnalyticDB for PostgreSQL | view, function, procedure, trigger, event |
| TDSQL (distributed) | YugabyteDB | view, function, procedure, trigger, event |
| AnalyticDB for MySQL | PostgreSQL | view, function, procedure, trigger, event |
| AnalyticDB for MySQL | CockroachDB | view, function, procedure, trigger, event |
| AnalyticDB for MySQL | Aurora DSQL | view, function, procedure, trigger, event |
| AnalyticDB for MySQL | GaussDB (openGauss) | view, function, procedure, trigger, event |
| AnalyticDB for MySQL | AnalyticDB for PostgreSQL | view, function, procedure, trigger, event |
| AnalyticDB for MySQL | YugabyteDB | view, function, procedure, trigger, event |
| AnalyticDB for PostgreSQL | MySQL | view, function, procedure, trigger |
| AnalyticDB for PostgreSQL | TiDB | view, function, procedure, trigger |
| AnalyticDB for PostgreSQL | OceanBase (MySQL mode) | view, function, procedure, trigger |
| AnalyticDB for PostgreSQL | PolarDB-X | view, function, procedure, trigger |
| AnalyticDB for PostgreSQL | TDSQL (distributed) | view, function, procedure, trigger |
| AnalyticDB for PostgreSQL | AnalyticDB for MySQL | view, function, procedure, trigger |
| AnalyticDB for PostgreSQL | SingleStore | view, function, procedure, trigger |
| YugabyteDB | MySQL | view, function, procedure, trigger |
| YugabyteDB | TiDB | view, function, procedure, trigger |
| YugabyteDB | OceanBase (MySQL mode) | view, function, procedure, trigger |
| YugabyteDB | PolarDB-X | view, function, procedure, trigger |
| YugabyteDB | TDSQL (distributed) | view, function, procedure, trigger |
| YugabyteDB | AnalyticDB for MySQL | view, function, procedure, trigger |
| YugabyteDB | SingleStore | view, function, procedure, trigger |
| SingleStore | PostgreSQL | view, function, procedure, trigger, event |
| SingleStore | CockroachDB | view, function, procedure, trigger, event |
| SingleStore | Aurora DSQL | view, function, procedure, trigger, event |
| SingleStore | GaussDB (openGauss) | view, function, procedure, trigger, event |
| SingleStore | AnalyticDB for PostgreSQL | view, function, procedure, trigger, event |
| SingleStore | YugabyteDB | view, function, procedure, trigger, event |
Both ways at once
A pair of hops, each side's changes carried to the other and migkit's own writes told apart, runs on: PostgreSQL, MySQL, MongoDB, Cassandra, Redis, DynamoDB, Kafka.
Managed services
A managed service is the engine it runs, reached at its address; migkit recognises these, and says the provider's own way of changing a setting on each.
| Provider | Service | Engine |
|---|---|---|
| AWS | Amazon RDS for PostgreSQL | PostgreSQL |
| AWS | Amazon RDS for MySQL | MySQL |
| AWS | Amazon RDS for MariaDB | MySQL |
| AWS | Amazon RDS for SQL Server | SQL Server |
| AWS | Amazon RDS for Oracle | Oracle Database |
| AWS | Amazon RDS for Db2 | Db2 |
| AWS | Amazon Aurora PostgreSQL | PostgreSQL |
| AWS | Amazon Aurora MySQL | MySQL |
| AWS | Amazon Aurora DSQL | Aurora DSQL |
| AWS | Amazon DocumentDB | MongoDB |
| AWS | Amazon DynamoDB | DynamoDB |
| AWS | Amazon ElastiCache (Valkey, Redis OSS) | Redis |
| AWS | Amazon MemoryDB | Redis |
| AWS | Amazon Keyspaces (for Apache Cassandra) | Keyspaces |
| AWS | Amazon Neptune | Neptune |
| AWS | Amazon Timestream for LiveAnalytics | Timestream |
| AWS | Amazon Timestream for InfluxDB | |
| AWS | Amazon Redshift | Redshift |
| AWS | Amazon OpenSearch Service | OpenSearch |
| AWS | Amazon MSK | Kafka |
| AWS | Amazon Kinesis Data Streams | a Kinesis stream |
| Azure | Azure SQL Database | SQL Server |
| Azure | Azure SQL Managed Instance | SQL Server |
| Azure | Azure Database for PostgreSQL flexible server | PostgreSQL |
| Azure | Azure Database for MySQL flexible server | MySQL |
| Azure | Azure Cosmos DB for NoSQL | Cosmos DB |
| Azure | Azure Cosmos DB for MongoDB (RU) | MongoDB |
| Azure | Azure DocumentDB (formerly Azure Cosmos DB for MongoDB vCore) | MongoDB |
| Azure | Azure Cosmos DB for Apache Cassandra | Cassandra |
| Azure | Azure Cosmos DB for PostgreSQL | PostgreSQL |
| Azure | Azure Cache for Redis | Redis |
| Azure | Azure Managed Redis | Redis |
| Azure | Azure Synapse Analytics (dedicated SQL pool) | Synapse dedicated SQL pool |
| Azure | Azure Event Hubs (Kafka endpoint) | Kafka |
| Google Cloud | Cloud SQL for PostgreSQL | PostgreSQL |
| Google Cloud | Cloud SQL for MySQL | MySQL |
| Google Cloud | Cloud SQL for SQL Server | SQL Server |
| Google Cloud | AlloyDB for PostgreSQL | PostgreSQL |
| Google Cloud | Spanner | Spanner |
| Google Cloud | Bigtable | Bigtable |
| Google Cloud | Firestore | Firestore |
| Google Cloud | Firestore in Datastore mode | Firestore in Datastore mode |
| Google Cloud | Memorystore for Redis | Redis |
| Google Cloud | Memorystore for Redis Cluster | Redis |
| Google Cloud | Memorystore for Valkey | Redis |
| Google Cloud | BigQuery | BigQuery |
| Google Cloud | Pub/Sub | a Pub/Sub topic |
| Tencent Cloud | TencentDB for PostgreSQL | PostgreSQL |
| Tencent Cloud | TencentDB for MySQL | MySQL |
| Tencent Cloud | TencentDB for MariaDB | MySQL |
| Tencent Cloud | TencentDB for SQL Server | SQL Server |
| Tencent Cloud | TencentDB for MongoDB | MongoDB |
| Tencent Cloud | TencentDB for Redis | Redis |
| Tencent Cloud | TDSQL-C for MySQL | MySQL |
| Tencent Cloud | TDSQL-C for PostgreSQL | PostgreSQL |
| Tencent Cloud | TDSQL (distributed) | TDSQL (distributed) |
| Tencent Cloud | Tencent Cloud CKafka | Kafka |
| Tencent Cloud | Tencent Cloud Elasticsearch Service | OpenSearch |
| Alibaba Cloud | ApsaraDB RDS for MySQL | MySQL |
| Alibaba Cloud | ApsaraDB RDS for PostgreSQL | PostgreSQL |
| Alibaba Cloud | ApsaraDB RDS for SQL Server | SQL Server |
| Alibaba Cloud | ApsaraDB RDS for MariaDB | MySQL |
| Alibaba Cloud | PolarDB for MySQL | MySQL |
| Alibaba Cloud | PolarDB for PostgreSQL | PostgreSQL |
| Alibaba Cloud | PolarDB-X | PolarDB-X |
| Alibaba Cloud | AnalyticDB for MySQL | AnalyticDB for MySQL |
| Alibaba Cloud | AnalyticDB for PostgreSQL | AnalyticDB for PostgreSQL |
| Alibaba Cloud | Tair (ApsaraDB for Redis) | Redis |
| Alibaba Cloud | ApsaraDB for MongoDB | MongoDB |
| Alibaba Cloud | Lindorm | Lindorm |
| Alibaba Cloud | ApsaraDB for OceanBase (MySQL mode) | OceanBase (MySQL mode) |
| Alibaba Cloud | ApsaraDB for OceanBase (Oracle mode) | OceanBase (Oracle mode) |
| Alibaba Cloud | Alibaba Cloud Elasticsearch | OpenSearch |
| Alibaba Cloud | ApsaraMQ for Kafka | Kafka |
| Huawei Cloud | Huawei Cloud RDS for MySQL | MySQL |
| Huawei Cloud | Huawei Cloud RDS for PostgreSQL | PostgreSQL |
| Huawei Cloud | Huawei Cloud RDS for SQL Server | SQL Server |
| Huawei Cloud | Huawei Cloud GaussDB | GaussDB |
| Huawei Cloud | Huawei Cloud TaurusDB (formerly GaussDB(for MySQL)) | MySQL |
| Huawei Cloud | Huawei Cloud Document Database Service (DDS) | MongoDB |
| Huawei Cloud | Huawei Cloud Distributed Cache Service (DCS) for Redis | Redis |
| Huawei Cloud | Huawei Cloud GeminiDB (DynamoDB, HBase and InfluxDB APIs) | |
| Huawei Cloud | Huawei Cloud GeminiDB (Cassandra API) | GeminiDB's Cassandra API |
| Huawei Cloud | Huawei Cloud GeminiDB (Redis API) | GeminiDB's Redis API |
| Huawei Cloud | Huawei Cloud GeminiDB (MongoDB API) | GeminiDB's MongoDB API |
| Huawei Cloud | Huawei Cloud Distributed Message Service (DMS) for Kafka | Kafka |
| Huawei Cloud | Huawei Cloud Search Service (CSS) | OpenSearch |
| Oracle Cloud | Oracle Autonomous Database | Oracle Database |
| Oracle Cloud | Oracle Base Database Service | Oracle Database |
| Oracle Cloud | MySQL HeatWave | MySQL |
| Oracle Cloud | OCI Database with PostgreSQL | PostgreSQL |
| Oracle Cloud | OCI Cache | Redis |
| Oracle Cloud | OCI Streaming (Kafka-compatible) | Kafka |
| IBM Cloud | IBM Cloud Databases for PostgreSQL | PostgreSQL |
| IBM Cloud | IBM Cloud Databases for MySQL | MySQL |
| IBM Cloud | IBM Cloud Databases for MongoDB | MongoDB |
| IBM Cloud | IBM Cloud Databases for Redis | Redis |
| IBM Cloud | IBM Cloud Databases for Elasticsearch | OpenSearch |
| IBM Cloud | IBM Event Streams | Kafka |
| IBM Cloud | IBM Db2 on Cloud (Db2 SaaS) | Db2 |
| IBM Cloud | IBM Db2 Warehouse on Cloud | Db2 |
| MongoDB Atlas | MongoDB Atlas | MongoDB |
| Redis Cloud | Redis Cloud | Redis |
| Aiven | Aiven for PostgreSQL | PostgreSQL |
| Aiven | Aiven for MySQL | MySQL |
| Aiven | Aiven for Apache Kafka | Kafka |
| Aiven | Aiven for OpenSearch | OpenSearch |
| Aiven | Aiven for Valkey | Redis |
| Aiven | Aiven for ClickHouse | ClickHouse |
| Aiven | Aiven for Apache Cassandra | Cassandra |
| Neon | Neon | PostgreSQL |
| Supabase | Supabase | PostgreSQL |
| PlanetScale | PlanetScale (MySQL, on Vitess) | MySQL |
| PlanetScale | PlanetScale for Postgres | PostgreSQL |
| Crunchy Bridge | Crunchy Bridge | PostgreSQL |
| Tiger Cloud (Timescale) | Tiger Cloud (Timescale) | PostgreSQL |
| CockroachDB Cloud | CockroachDB Cloud | CockroachDB |
| YugabyteDB Aeon | YugabyteDB Aeon | YugabyteDB |
| SingleStore Helios | SingleStore Helios | SingleStore |
| Confluent Cloud | Confluent Cloud | Kafka |
| Teradata VantageCloud | Teradata VantageCloud Lake | Teradata |
Each engine
PostgreSQL
postgres, also written alloydb, aurora-postgres, pgsql, postgresql - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
MySQL
mysql, also written aurora-mysql, mariadb, percona, tdsql, tdsql-c - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
MongoDB
mongodb, also written cosmosdb-mongo, documentdb - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: collections have no sequences or auto-increment
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
SQL Server
mssql, also written azure-sql, sqlserver - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Redis
redis, also written valkey - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: keys have no sequences
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
Kafka
kafka - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: offsets are assigned by the broker, not carried
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
SQLite
sqlite - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; moving a whole database in bulk; copying table by table, resumably; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Does not apply:
- keeping the target following the source: a SQLite database is a file with no change log that can be read without writing to it, so it moves offline
- proving the target has caught up before cutover: a SQLite database is a file with no change log that can be read without writing to it, so it moves offline
- telling a difference still arriving from one that is wrong: a SQLite database is a file with no change log that can be read without writing to it, so it moves offline
- verifying only what changed: a SQLite database is a file with no change log that can be read without writing to it, so it moves offline
- carrying users and their grants: a SQLite database has no users
Parquet files
parquet - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; moving a whole database in bulk; copying table by table, resumably; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: Parquet files are data at rest, with no server, settings or change log of their own
- comparing server settings: Parquet files are data at rest, with no server, settings or change log of their own
- keeping the target following the source: Parquet files are data at rest, with no server, settings or change log of their own
- proving the target has caught up before cutover: Parquet files are data at rest, with no server, settings or change log of their own
- telling a difference still arriving from one that is wrong: Parquet files are data at rest, with no server, settings or change log of their own
- verifying only what changed: Parquet files are data at rest, with no server, settings or change log of their own
- carrying users and their grants: who may read the files is the storage's own access control, not the table's
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
ClickHouse
clickhouse - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: ClickHouse has no sequences or auto-increment to carry
- refreshing the target's statistics after a load: a MergeTree reads by its sorting key and keeps no optimiser statistics to refresh
DynamoDB
dynamodb - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: DynamoDB has no sequences or auto-increment to carry
- carrying users and their grants: who may read a table is IAM's, not the table's
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
Oracle Database
oracle - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Db2 (Linux, UNIX, Windows)
db2 - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Db2 for i
db2i, also written as400, db2-for-i, ibmi, iseries - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- refreshing the target's statistics after a load: IBM i's query engine gathers and keeps its own statistics as the tables change
Db2 for z/OS
db2zos, also written db2-zos, db2z - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; moving a whole database in bulk; copying table by table, resumably; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Does not apply:
- keeping the target following the source: Db2 for z/OS hands its log out only to licensed code - a load module on the host reading it, or the change tables IBM's replication product fills - and no open reader of either exists, so a z/OS source moves by copies, each proved by check
- proving the target has caught up before cutover: Db2 for z/OS hands its log out only to licensed code - a load module on the host reading it, or the change tables IBM's replication product fills - and no open reader of either exists, so a z/OS source moves by copies, each proved by check
- telling a difference still arriving from one that is wrong: Db2 for z/OS hands its log out only to licensed code - a load module on the host reading it, or the change tables IBM's replication product fills - and no open reader of either exists, so a z/OS source moves by copies, each proved by check
- verifying only what changed: Db2 for z/OS hands its log out only to licensed code - a load module on the host reading it, or the change tables IBM's replication product fills - and no open reader of either exists, so a z/OS source moves by copies, each proved by check
SAP ASE
ase, also written sap-ase, sybase - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; moving a whole database in bulk; copying table by table, resumably; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Not yet:
- keeping the target following the source - meanwhile, stop writes on the source, move it once, and prove it with migkit check before cutover
Teradata
teradata, also written td, vantage - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; copying table by table, resumably; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load.
Does not apply:
- carrying sequences and auto-increment values: Teradata has no sequences, and an identity column's next number is the system's own, set on no table that exists
- comparing server settings: Teradata's system settings are read by its own utility on a node, not through SQL
- moving a whole database in bulk: Teradata's whole-system copy is its backup and restore, licensed apart from the database; tables are copied in its parallel sessions (table-copy)
- keeping the target following the source: Teradata keeps no change log a client can read, so a Teradata source moves by copies, each proved by check
- proving the target has caught up before cutover: Teradata keeps no change log a client can read, so a Teradata source moves by copies, each proved by check
- telling a difference still arriving from one that is wrong: Teradata keeps no change log a client can read, so a Teradata source moves by copies, each proved by check
- verifying only what changed: Teradata keeps no change log a client can read, so a Teradata source moves by copies, each proved by check
- snapshotting the target so a cutover can be rolled back: Teradata keeps no snapshot a client can take; a copy of the target is its backup tool's
OpenSearch
opensearch, also written elasticsearch - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: an index has no sequences or auto-increment to carry
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
Cassandra
cassandra, also written scylla, scylladb - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: Cassandra has no sequences or auto-increment to carry
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
Amazon Redshift
redshift - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; copying table by table, resumably; carrying users and their grants; noticing a move that moved nothing.
Does not apply:
- moving a whole database in bulk: a warehouse has no whole-database dump and load a client runs: its bulk path is its own stage load, which table-copy runs table by table
- keeping the target following the source: Redshift keeps no change history a client reads: a Redshift source is copied, not followed, and a Redshift target follows by its landing's position
- proving the target has caught up before cutover: nothing follows a Redshift source, so there is no catch-up to wait for; a writer is held to its number in migkit's own table, serializable
- telling a difference still arriving from one that is wrong: nothing follows a Redshift source, so nothing is still arriving: every difference is one
- verifying only what changed: Redshift records no changed rows a client reads: the full check compares every row
- refreshing the target's statistics after a load: Redshift analyzes a table itself after a load (its automatic analyze)
- snapshotting the target so a cutover can be rolled back: a Redshift snapshot is the cluster's, taken by the service, not by a client: the rollback point is the service's own snapshot
Snowflake
snowflake - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- moving a whole database in bulk: a warehouse has no whole-database dump and load a client runs: its bulk path is its own stage load, which table-copy runs table by table
- refreshing the target's statistics after a load: Snowflake keeps each table's statistics itself, in its partitions' metadata written with each load
BigQuery
bigquery - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: BigQuery has no sequences or auto-increment to carry
- comparing server settings: BigQuery has no server settings to compare: what a dataset and a table are kept under are their options, which the deep checks compare
- refreshing the target's statistics after a load: BigQuery keeps its tables' statistics itself as they change: there is nothing to refresh
Kinesis Data Streams
kinesis - a stream a pair delivers its changes into.
Does not apply:
- comparing the schema: Kinesis is where a pair delivers its changes (engine: hetero, target_engine: kinesis), not a hop's engine: it holds a stream of changes, not a copy to compare
- comparing row counts: Kinesis is where a pair delivers its changes (engine: hetero, target_engine: kinesis), not a hop's engine: it holds a stream of changes, not a copy to compare
- comparing the data itself: Kinesis is where a pair delivers its changes (engine: hetero, target_engine: kinesis), not a hop's engine: it holds a stream of changes, not a copy to compare
- the deep checks: Kinesis is where a pair delivers its changes (engine: hetero, target_engine: kinesis), not a hop's engine: it holds a stream of changes, not a copy to compare
- carrying sequences and auto-increment values: Kinesis is where a pair delivers its changes (engine: hetero, target_engine: kinesis), not a hop's engine: it holds a stream of changes, not a copy to compare
- comparing server settings: Kinesis is where a pair delivers its changes (engine: hetero, target_engine: kinesis), not a hop's engine: it holds a stream of changes, not a copy to compare
- moving a whole database in bulk: Kinesis is where a pair delivers its changes (engine: hetero, target_engine: kinesis), not a hop's engine: it holds a stream of changes, not a copy to compare
- copying table by table, resumably: Kinesis is where a pair delivers its changes (engine: hetero, target_engine: kinesis), not a hop's engine: it holds a stream of changes, not a copy to compare
- keeping the target following the source: Kinesis is where a pair delivers its changes (engine: hetero, target_engine: kinesis), not a hop's engine: it holds a stream of changes, not a copy to compare
- proving the target has caught up before cutover: Kinesis is where a pair delivers its changes (engine: hetero, target_engine: kinesis), not a hop's engine: it holds a stream of changes, not a copy to compare
- telling a difference still arriving from one that is wrong: Kinesis is where a pair delivers its changes (engine: hetero, target_engine: kinesis), not a hop's engine: it holds a stream of changes, not a copy to compare
- verifying only what changed: Kinesis is where a pair delivers its changes (engine: hetero, target_engine: kinesis), not a hop's engine: it holds a stream of changes, not a copy to compare
- carrying users and their grants: Kinesis is where a pair delivers its changes (engine: hetero, target_engine: kinesis), not a hop's engine: it holds a stream of changes, not a copy to compare
- noticing a move that moved nothing: Kinesis is where a pair delivers its changes (engine: hetero, target_engine: kinesis), not a hop's engine: it holds a stream of changes, not a copy to compare
- refreshing the target's statistics after a load: Kinesis is where a pair delivers its changes (engine: hetero, target_engine: kinesis), not a hop's engine: it holds a stream of changes, not a copy to compare
- snapshotting the target so a cutover can be rolled back: Kinesis is where a pair delivers its changes (engine: hetero, target_engine: kinesis), not a hop's engine: it holds a stream of changes, not a copy to compare
Google Pub/Sub
pubsub - a stream a pair delivers its changes into.
Does not apply:
- comparing the schema: Pub/Sub is where a pair delivers its changes (engine: hetero, target_engine: pubsub), not a hop's engine: it holds a stream of changes, not a copy to compare
- comparing row counts: Pub/Sub is where a pair delivers its changes (engine: hetero, target_engine: pubsub), not a hop's engine: it holds a stream of changes, not a copy to compare
- comparing the data itself: Pub/Sub is where a pair delivers its changes (engine: hetero, target_engine: pubsub), not a hop's engine: it holds a stream of changes, not a copy to compare
- the deep checks: Pub/Sub is where a pair delivers its changes (engine: hetero, target_engine: pubsub), not a hop's engine: it holds a stream of changes, not a copy to compare
- carrying sequences and auto-increment values: Pub/Sub is where a pair delivers its changes (engine: hetero, target_engine: pubsub), not a hop's engine: it holds a stream of changes, not a copy to compare
- comparing server settings: Pub/Sub is where a pair delivers its changes (engine: hetero, target_engine: pubsub), not a hop's engine: it holds a stream of changes, not a copy to compare
- moving a whole database in bulk: Pub/Sub is where a pair delivers its changes (engine: hetero, target_engine: pubsub), not a hop's engine: it holds a stream of changes, not a copy to compare
- copying table by table, resumably: Pub/Sub is where a pair delivers its changes (engine: hetero, target_engine: pubsub), not a hop's engine: it holds a stream of changes, not a copy to compare
- keeping the target following the source: Pub/Sub is where a pair delivers its changes (engine: hetero, target_engine: pubsub), not a hop's engine: it holds a stream of changes, not a copy to compare
- proving the target has caught up before cutover: Pub/Sub is where a pair delivers its changes (engine: hetero, target_engine: pubsub), not a hop's engine: it holds a stream of changes, not a copy to compare
- telling a difference still arriving from one that is wrong: Pub/Sub is where a pair delivers its changes (engine: hetero, target_engine: pubsub), not a hop's engine: it holds a stream of changes, not a copy to compare
- verifying only what changed: Pub/Sub is where a pair delivers its changes (engine: hetero, target_engine: pubsub), not a hop's engine: it holds a stream of changes, not a copy to compare
- carrying users and their grants: Pub/Sub is where a pair delivers its changes (engine: hetero, target_engine: pubsub), not a hop's engine: it holds a stream of changes, not a copy to compare
- noticing a move that moved nothing: Pub/Sub is where a pair delivers its changes (engine: hetero, target_engine: pubsub), not a hop's engine: it holds a stream of changes, not a copy to compare
- refreshing the target's statistics after a load: Pub/Sub is where a pair delivers its changes (engine: hetero, target_engine: pubsub), not a hop's engine: it holds a stream of changes, not a copy to compare
- snapshotting the target so a cutover can be rolled back: Pub/Sub is where a pair delivers its changes (engine: hetero, target_engine: pubsub), not a hop's engine: it holds a stream of changes, not a copy to compare
DuckDB
duckdb - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; moving a whole database in bulk; copying table by table, resumably; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- comparing server settings: a DuckDB file keeps no settings: they are the process's that opens it
- keeping the target following the source: a SQLite database is a file with no change log that can be read without writing to it, so it moves offline
- proving the target has caught up before cutover: a SQLite database is a file with no change log that can be read without writing to it, so it moves offline
- telling a difference still arriving from one that is wrong: a SQLite database is a file with no change log that can be read without writing to it, so it moves offline
- verifying only what changed: a SQLite database is a file with no change log that can be read without writing to it, so it moves offline
- carrying users and their grants: a DuckDB database has no users
- refreshing the target's statistics after a load: DuckDB keeps its statistics as it writes
Files described by COBOL copybooks
copybook, also written cobol, vsam - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; moving a whole database in bulk; copying table by table, resumably; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: a file described by a copybook is an extract at rest, with no server, settings or change log; what changed since the last extract is found by comparing the two, which check does
- comparing server settings: a file described by a copybook is an extract at rest, with no server, settings or change log; what changed since the last extract is found by comparing the two, which check does
- keeping the target following the source: a file described by a copybook is an extract at rest, with no server, settings or change log; what changed since the last extract is found by comparing the two, which check does
- proving the target has caught up before cutover: a file described by a copybook is an extract at rest, with no server, settings or change log; what changed since the last extract is found by comparing the two, which check does
- telling a difference still arriving from one that is wrong: a file described by a copybook is an extract at rest, with no server, settings or change log; what changed since the last extract is found by comparing the two, which check does
- verifying only what changed: a file described by a copybook is an extract at rest, with no server, settings or change log; what changed since the last extract is found by comparing the two, which check does
- carrying users and their grants: who may read a dataset is the security manager's or the file system's, not the file's
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
Iceberg tables
iceberg - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; copying table by table, resumably; noticing a move that moved nothing.
Does not apply:
- carrying sequences and auto-increment values: an Iceberg or Delta table is files and a log of commits, with no server, settings or sequences of its own
- comparing server settings: an Iceberg or Delta table is files and a log of commits, with no server, settings or sequences of its own
- carrying users and their grants: who may read the table is the catalog's and the storage's own access control, not the table's
- refreshing the target's statistics after a load: the table's statistics are its own files' footers and manifests, written with each commit
Not yet:
- the deep checks - meanwhile, run migkit check, whose other checks still apply
- moving a whole database in bulk - meanwhile, move it table by table, or restore a backup of the source
- keeping the target following the source - meanwhile, stop writes on the source, move it once, and prove it with migkit check before cutover
- proving the target has caught up before cutover - meanwhile, stop writes on the source and let migkit check prove the target before cutover
- telling a difference still arriving from one that is wrong - meanwhile, stop writes on the source before the final check, so nothing is still arriving when it looks
- verifying only what changed - meanwhile, run migkit check, which compares everything rather than only what changed
- snapshotting the target so a cutover can be rolled back - meanwhile, take a backup of the target yourself before cutover
Delta tables
delta, also written delta-lake - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; copying table by table, resumably; noticing a move that moved nothing.
Does not apply:
- carrying sequences and auto-increment values: an Iceberg or Delta table is files and a log of commits, with no server, settings or sequences of its own
- comparing server settings: an Iceberg or Delta table is files and a log of commits, with no server, settings or sequences of its own
- carrying users and their grants: who may read the table is the catalog's and the storage's own access control, not the table's
- refreshing the target's statistics after a load: the table's statistics are its own files' footers and manifests, written with each commit
Not yet:
- the deep checks - meanwhile, run migkit check, whose other checks still apply
- moving a whole database in bulk - meanwhile, move it table by table, or restore a backup of the source
- keeping the target following the source - meanwhile, stop writes on the source, move it once, and prove it with migkit check before cutover
- proving the target has caught up before cutover - meanwhile, stop writes on the source and let migkit check prove the target before cutover
- telling a difference still arriving from one that is wrong - meanwhile, stop writes on the source before the final check, so nothing is still arriving when it looks
- verifying only what changed - meanwhile, run migkit check, which compares everything rather than only what changed
- snapshotting the target so a cutover can be rolled back - meanwhile, take a backup of the target yourself before cutover
TiDB
tidb - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
CockroachDB
cockroachdb, also written cockroach, crdb - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- refreshing the target's statistics after a load: CockroachDB collects its tables' statistics itself as they change
Aurora DSQL
dsql, also written aurora-dsql - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Does not apply:
- comparing server settings: a DSQL cluster keeps no settings of its operator's to compare
- moving a whole database in bulk: DSQL takes 3,000 rows a transaction whatever writes them and has no bulk load: the table copier writes within it
- carrying users and their grants: who may sign in to a DSQL cluster is IAM's: a role is given by an IAM policy, not made in the database
Couchbase
couchbase - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: a bucket has no sequences or auto-increment to carry
Cosmos DB for NoSQL
cosmosdb, also written azure-cosmos, cosmos, cosmosdb-nosql - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: a container has no sequences or auto-increment to carry
- carrying users and their grants: who may read a container is the account's role assignments, not the container's
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
Google Cloud Spanner
spanner, also written cloud-spanner, spanner-pg - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Google Cloud Bigtable
bigtable, also written cloud-bigtable - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: a row is kept by the key the application gives it, not one a sequence hands out: there is no sequence to carry
- carrying users and their grants: who may read a table is the cloud's access policies', outside the database
- refreshing the target's statistics after a load: a Bigtable table is read by its row keys and keeps no optimiser statistics
Firestore (native mode)
firestore, also written firestore-native - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: a document is kept by the key the application gives it, not one a sequence hands out: there is no sequence to carry
- carrying users and their grants: who may read a database is IAM's and its security rules', not the collection's
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
Firestore in Datastore mode
datastore, also written cloud-datastore, firestore-datastore - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; moving a whole database in bulk; copying table by table, resumably; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: an entity's numeric id is allocated by the service and carried as it is: there is no sequence to carry
- keeping the target following the source: Datastore mode has no change feed and no listener: nothing a client can follow
- proving the target has caught up before cutover: Datastore mode has no change feed and no listener: nothing a client can follow
- telling a difference still arriving from one that is wrong: Datastore mode has no change feed and no listener: nothing a client can follow
- verifying only what changed: Datastore mode has no change feed and no listener: nothing a client can follow
- carrying users and their grants: who may read a kind is the cloud's access policies', outside the database
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
GaussDB (openGauss)
gaussdb, also written gaussdb-opengauss, opengauss - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
OceanBase (MySQL mode)
oceanbase, also written oceanbase-mysql - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Not yet:
- moving a whole database in bulk - meanwhile, move it table by table, or restore a backup of the source
OceanBase (Oracle mode)
oceanbase-oracle - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; copying table by table, resumably; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Not yet:
- moving a whole database in bulk - meanwhile, move it table by table, or restore a backup of the source
- keeping the target following the source - meanwhile, stop writes on the source, move it once, and prove it with migkit check before cutover
- proving the target has caught up before cutover - meanwhile, stop writes on the source and let migkit check prove the target before cutover
- telling a difference still arriving from one that is wrong - meanwhile, stop writes on the source before the final check, so nothing is still arriving when it looks
- verifying only what changed - meanwhile, run migkit check, which compares everything rather than only what changed
PolarDB-X
polardbx, also written drds, polardb-x - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Not yet:
- moving a whole database in bulk - meanwhile, move it table by table, or restore a backup of the source
TDSQL (distributed)
dcdb, also written tdsql-distributed, tdsql-shard - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Not yet:
- moving a whole database in bulk - meanwhile, move it table by table, or restore a backup of the source
Amazon Keyspaces
keyspaces, also written amazon-keyspaces, aws-keyspaces - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: Cassandra has no sequences or auto-increment to carry
- carrying users and their grants: who may read a table is the cloud's access policies', outside the database
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
Amazon Neptune
neptune, also written amazon-neptune - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: a vertex or edge is kept by the key the application gives it, not one a sequence hands out: there is no sequence to carry
- carrying users and their grants: who may read the graph is the cloud's access policies', outside the database
Amazon Timestream
timestream, also written amazon-timestream - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; comparing server settings; moving a whole database in bulk; copying table by table, resumably; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- the deep checks: a record is its dimensions, time and measures, which the data check compares whole; there are no indexes or constraints to hold
- carrying sequences and auto-increment values: a record is kept by its dimensions and its time: there is no sequence to carry
- keeping the target following the source: Timestream keeps no change log, and a record written late or at a higher version lands in the past, where no watermark finds it
- proving the target has caught up before cutover: Timestream keeps no change log, and a record written late or at a higher version lands in the past, where no watermark finds it
- telling a difference still arriving from one that is wrong: Timestream keeps no change log, and a record written late or at a higher version lands in the past, where no watermark finds it
- verifying only what changed: Timestream keeps no change log, and a record written late or at a higher version lands in the past, where no watermark finds it
- carrying users and their grants: who may read a table is the cloud's access policies', outside the database
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
Lindorm
lindorm - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; copying table by table, resumably; carrying users and their grants; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: a row is kept by the key the application gives it, not one a sequence hands out: there is no sequence to carry
- moving a whole database in bulk: Lindorm's own bulk paths - its tunnel service, its bulk load - are run from its console, not by a client: its tables go by the table copier
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
Not yet:
- keeping the target following the source - meanwhile, stop writes on the source, move it once, and prove it with migkit check before cutover
- proving the target has caught up before cutover - meanwhile, stop writes on the source and let migkit check prove the target before cutover
- telling a difference still arriving from one that is wrong - meanwhile, stop writes on the source before the final check, so nothing is still arriving when it looks
- verifying only what changed - meanwhile, run migkit check, which compares everything rather than only what changed
GeminiDB (Cassandra API)
geminidb-cassandra, also written geminidb - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; moving a whole database in bulk; copying table by table, resumably; carrying users and their grants; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: Cassandra has no sequences or auto-increment to carry
- keeping the target following the source: GeminiDB's Cassandra API keeps no change log a client can read
- proving the target has caught up before cutover: GeminiDB's Cassandra API keeps no change log a client can read
- telling a difference still arriving from one that is wrong: GeminiDB's Cassandra API keeps no change log a client can read
- verifying only what changed: GeminiDB's Cassandra API keeps no change log a client can read
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
GeminiDB (Redis API)
geminidb-redis - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; copying table by table, resumably; keeping the target following the source; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: keys have no sequences
- moving a whole database in bulk: the Redis copy programs read a replica's stream, which GeminiDB serves none of, or are licensed into Redis's own products alone: its keys go by migkit's own copier
- proving the target has caught up before cutover: a GeminiDB instance neither is nor serves a replica, and keeps no change stream a client can read
- telling a difference still arriving from one that is wrong: a GeminiDB instance neither is nor serves a replica, and keeps no change stream a client can read
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
GeminiDB (MongoDB API)
geminidb-mongo - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: collections have no sequences or auto-increment
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
Qdrant
qdrant - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; copying table by table, resumably; noticing a move that moved nothing.
Does not apply:
- carrying sequences and auto-increment values: a point is kept by the key the application gives it, not one a sequence hands out: there is no sequence to carry
- keeping the target following the source: Qdrant keeps no change log a client can read: a move is copied and checked again before the switch
- proving the target has caught up before cutover: Qdrant keeps no change log a client can read: a move is copied and checked again before the switch
- telling a difference still arriving from one that is wrong: Qdrant keeps no change log a client can read: a move is copied and checked again before the switch
- verifying only what changed: Qdrant keeps no change log a client can read: a move is copied and checked again before the switch
- carrying users and their grants: who may read a collection is the server's API keys, kept in its configuration
- refreshing the target's statistics after a load: the engine keeps no optimiser statistics to refresh
Not yet:
- moving a whole database in bulk - meanwhile, move it table by table, or restore a backup of the source
- snapshotting the target so a cutover can be rolled back - meanwhile, take a backup of the target yourself before cutover
Neo4j
neo4j, also written neo4j-aura - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load.
Does not apply:
- carrying sequences and auto-increment values: a node or relationship is kept by the key the application gives it, not one a sequence hands out: there is no sequence to carry
Not yet:
- moving a whole database in bulk - meanwhile, move it table by table, or restore a backup of the source
- telling a difference still arriving from one that is wrong - meanwhile, stop writes on the source before the final check, so nothing is still arriving when it looks
- snapshotting the target so a cutover can be rolled back - meanwhile, take a backup of the target yourself before cutover
AnalyticDB for MySQL
analyticdb-mysql, also written adb-mysql - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Does not apply:
- moving a whole database in bulk: AnalyticDB for MySQL loads a whole table through its own loader from a landing store, ranked against the table copier table by table (
decide.landing_load): no dump and load of a whole database reaches it
AnalyticDB for PostgreSQL
analyticdb-postgres, also written adb-pg, analyticdb-pg - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing.
Does not apply:
- moving a whole database in bulk: AnalyticDB for PostgreSQL loads a whole table through its own loader from a landing store, ranked against the table copier table by table (
decide.landing_load): no dump and load of a whole database reaches it
Not yet:
- comparing server settings - meanwhile, compare the server settings by hand
- refreshing the target's statistics after a load - meanwhile, refresh the target's statistics by hand after the load
- snapshotting the target so a cutover can be rolled back - meanwhile, take a backup of the target yourself before cutover
Azure Synapse (dedicated SQL pool)
synapse, also written azure-synapse - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; comparing server settings; copying table by table, resumably; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Does not apply:
- carrying sequences and auto-increment values: a dedicated SQL pool has no sequences to carry; identity columns are set on table creation
- moving a whole database in bulk: Synapse's dedicated SQL pool loads a whole table through its own loader from a landing store, ranked against the table copier table by table (
decide.landing_load): no dump and load of a whole database reaches it - keeping the target following the source: a dedicated SQL pool keeps no change data capture or change tracking a client reads, and is a pair's warehouse target
- proving the target has caught up before cutover: a dedicated SQL pool keeps no change data capture or change tracking a client reads, and is a pair's warehouse target
- telling a difference still arriving from one that is wrong: a dedicated SQL pool keeps no change data capture or change tracking a client reads, and is a pair's warehouse target
- verifying only what changed: a dedicated SQL pool keeps no change data capture or change tracking a client reads, and is a pair's warehouse target
YugabyteDB
yugabytedb, also written ysql, yugabyte - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing.
Does not apply:
- moving a whole database in bulk: YugabyteDB takes rows through PostgreSQL's COPY, which the table copier already writes by key in batches: no dump and load of its own reaches a client
Not yet:
- comparing server settings - meanwhile, compare the server settings by hand
- refreshing the target's statistics after a load - meanwhile, refresh the target's statistics by hand after the load
- snapshotting the target so a cutover can be rolled back - meanwhile, take a backup of the target yourself before cutover
SingleStore
singlestore, also written memsql - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Does not apply:
- moving a whole database in bulk: SingleStore loads a whole table through its own loader from a landing store, ranked against the table copier table by table (
decide.landing_load): no dump and load of a whole database reaches it
Netezza
netezza, also written pure-data-netezza - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; noticing a move that moved nothing.
Does not apply:
- carrying sequences and auto-increment values: the generic engine reaches a server only through a standard driver, which has no statement for a sequence or auto-increment counter; a server whose one migkit reads is a hop of its own engine
- comparing server settings: the generic engine reaches a server only through a standard driver, which has no statement for a server setting; a server whose one migkit reads is a hop of its own engine
- moving a whole database in bulk: a whole database moved in bulk is the server's own dump and load, which a driver does not reach; a table goes in by the dialect's own bulk way where the climb finds one that holds (
decide.dialect_bulk), table by table - carrying users and their grants: the generic engine reaches a server only through a standard driver, which has no statement for an account or its grants; a server whose one migkit reads is a hop of its own engine
- refreshing the target's statistics after a load: the generic engine reaches a server only through a standard driver, which has no statement for a refresh of the optimiser's statistics; a server whose one migkit reads is a hop of its own engine
- snapshotting the target so a cutover can be rolled back: the generic engine reaches a server only through a standard driver, which has no statement for a snapshot of the server; a server whose one migkit reads is a hop of its own engine
SAP HANA
saphana, also written hana, sap-hana - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; noticing a move that moved nothing.
Does not apply:
- carrying sequences and auto-increment values: the generic engine reaches a server only through a standard driver, which has no statement for a sequence or auto-increment counter; a server whose one migkit reads is a hop of its own engine
- comparing server settings: the generic engine reaches a server only through a standard driver, which has no statement for a server setting; a server whose one migkit reads is a hop of its own engine
- moving a whole database in bulk: a whole database moved in bulk is the server's own dump and load, which a driver does not reach; a table goes in by the dialect's own bulk way where the climb finds one that holds (
decide.dialect_bulk), table by table - carrying users and their grants: the generic engine reaches a server only through a standard driver, which has no statement for an account or its grants; a server whose one migkit reads is a hop of its own engine
- refreshing the target's statistics after a load: the generic engine reaches a server only through a standard driver, which has no statement for a refresh of the optimiser's statistics; a server whose one migkit reads is a hop of its own engine
- snapshotting the target so a cutover can be rolled back: the generic engine reaches a server only through a standard driver, which has no statement for a snapshot of the server; a server whose one migkit reads is a hop of its own engine
Informix
informix - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; noticing a move that moved nothing.
Does not apply:
- carrying sequences and auto-increment values: the generic engine reaches a server only through a standard driver, which has no statement for a sequence or auto-increment counter; a server whose one migkit reads is a hop of its own engine
- comparing server settings: the generic engine reaches a server only through a standard driver, which has no statement for a server setting; a server whose one migkit reads is a hop of its own engine
- moving a whole database in bulk: a whole database moved in bulk is the server's own dump and load, which a driver does not reach; a table goes in by the dialect's own bulk way where the climb finds one that holds (
decide.dialect_bulk), table by table - carrying users and their grants: the generic engine reaches a server only through a standard driver, which has no statement for an account or its grants; a server whose one migkit reads is a hop of its own engine
- refreshing the target's statistics after a load: the generic engine reaches a server only through a standard driver, which has no statement for a refresh of the optimiser's statistics; a server whose one migkit reads is a hop of its own engine
- snapshotting the target so a cutover can be rolled back: the generic engine reaches a server only through a standard driver, which has no statement for a snapshot of the server; a server whose one migkit reads is a hop of its own engine
Progress OpenEdge
progress, also written openedge, progress-openedge - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; noticing a move that moved nothing.
Does not apply:
- carrying sequences and auto-increment values: the generic engine reaches a server only through a standard driver, which has no statement for a sequence or auto-increment counter; a server whose one migkit reads is a hop of its own engine
- comparing server settings: the generic engine reaches a server only through a standard driver, which has no statement for a server setting; a server whose one migkit reads is a hop of its own engine
- moving a whole database in bulk: a whole database moved in bulk is the server's own dump and load, which a driver does not reach; a table goes in by the dialect's own bulk way where the climb finds one that holds (
decide.dialect_bulk), table by table - carrying users and their grants: the generic engine reaches a server only through a standard driver, which has no statement for an account or its grants; a server whose one migkit reads is a hop of its own engine
- refreshing the target's statistics after a load: the generic engine reaches a server only through a standard driver, which has no statement for a refresh of the optimiser's statistics; a server whose one migkit reads is a hop of its own engine
- snapshotting the target so a cutover can be rolled back: the generic engine reaches a server only through a standard driver, which has no statement for a snapshot of the server; a server whose one migkit reads is a hop of its own engine
Vertica
vertica - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; noticing a move that moved nothing.
Does not apply:
- carrying sequences and auto-increment values: the generic engine reaches a server only through a standard driver, which has no statement for a sequence or auto-increment counter; a server whose one migkit reads is a hop of its own engine
- comparing server settings: the generic engine reaches a server only through a standard driver, which has no statement for a server setting; a server whose one migkit reads is a hop of its own engine
- moving a whole database in bulk: a whole database moved in bulk is the server's own dump and load, which a driver does not reach; a table goes in by the dialect's own bulk way where the climb finds one that holds (
decide.dialect_bulk), table by table - carrying users and their grants: the generic engine reaches a server only through a standard driver, which has no statement for an account or its grants; a server whose one migkit reads is a hop of its own engine
- refreshing the target's statistics after a load: the generic engine reaches a server only through a standard driver, which has no statement for a refresh of the optimiser's statistics; a server whose one migkit reads is a hop of its own engine
- snapshotting the target so a cutover can be rolled back: the generic engine reaches a server only through a standard driver, which has no statement for a snapshot of the server; a server whose one migkit reads is a hop of its own engine
Exasol
exasol - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; noticing a move that moved nothing.
Does not apply:
- carrying sequences and auto-increment values: the generic engine reaches a server only through a standard driver, which has no statement for a sequence or auto-increment counter; a server whose one migkit reads is a hop of its own engine
- comparing server settings: the generic engine reaches a server only through a standard driver, which has no statement for a server setting; a server whose one migkit reads is a hop of its own engine
- moving a whole database in bulk: a whole database moved in bulk is the server's own dump and load, which a driver does not reach; a table goes in by the dialect's own bulk way where the climb finds one that holds (
decide.dialect_bulk), table by table - carrying users and their grants: the generic engine reaches a server only through a standard driver, which has no statement for an account or its grants; a server whose one migkit reads is a hop of its own engine
- refreshing the target's statistics after a load: the generic engine reaches a server only through a standard driver, which has no statement for a refresh of the optimiser's statistics; a server whose one migkit reads is a hop of its own engine
- snapshotting the target so a cutover can be rolled back: the generic engine reaches a server only through a standard driver, which has no statement for a snapshot of the server; a server whose one migkit reads is a hop of its own engine
Databricks SQL
databricks, also written databricks-sql - either side.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; noticing a move that moved nothing.
Does not apply:
- carrying sequences and auto-increment values: the generic engine reaches a server only through a standard driver, which has no statement for a sequence or auto-increment counter; a server whose one migkit reads is a hop of its own engine
- comparing server settings: the generic engine reaches a server only through a standard driver, which has no statement for a server setting; a server whose one migkit reads is a hop of its own engine
- moving a whole database in bulk: a whole database moved in bulk is the server's own dump and load, which a driver does not reach; a table goes in by the dialect's own bulk way where the climb finds one that holds (
decide.dialect_bulk), table by table - carrying users and their grants: the generic engine reaches a server only through a standard driver, which has no statement for an account or its grants; a server whose one migkit reads is a hop of its own engine
- refreshing the target's statistics after a load: the generic engine reaches a server only through a standard driver, which has no statement for a refresh of the optimiser's statistics; a server whose one migkit reads is a hop of its own engine
- snapshotting the target so a cutover can be rolled back: the generic engine reaches a server only through a standard driver, which has no statement for a snapshot of the server; a server whose one migkit reads is a hop of its own engine
SAP application data (ODP)
sapodp, also written odp, sap-odp - a source, read and never written.
It does: comparing the schema; comparing row counts; comparing the data itself.
Does not apply:
- the deep checks: an SAP system's application data holds no state of a load to look into: it is read through its own interface
- carrying sequences and auto-increment values: an SAP system's application data is read, never written: it is a pair's source (engine: hetero, source_engine: sapodp), and a hop between two of them compares them
- comparing server settings: an SAP system's application data is read, never written: it is a pair's source (engine: hetero, source_engine: sapodp), and a hop between two of them compares them
- moving a whole database in bulk: an SAP system's application data is read, never written: it is a pair's source (engine: hetero, source_engine: sapodp), and a hop between two of them compares them
- copying table by table, resumably: an SAP system's application data is read, never written: it is a pair's source (engine: hetero, source_engine: sapodp), and a hop between two of them compares them
- keeping the target following the source: an SAP system's application data is followed as a pair's source, its changes carried into another engine (engine: hetero, source_engine: sapodp); two of them are compared, never one written
- proving the target has caught up before cutover: an SAP system's application data is followed as a pair's source, its changes carried into another engine (engine: hetero, source_engine: sapodp); two of them are compared, never one written
- telling a difference still arriving from one that is wrong: an SAP system's application data is followed as a pair's source, its changes carried into another engine (engine: hetero, source_engine: sapodp); two of them are compared, never one written
- verifying only what changed: an SAP system's application data is followed as a pair's source, its changes carried into another engine (engine: hetero, source_engine: sapodp); two of them are compared, never one written
- carrying users and their grants: SAP authorisations are the system's roles (PFCG), not carried
- noticing a move that moved nothing: an SAP system's application data is read, never written: it is a pair's source (engine: hetero, source_engine: sapodp), and a hop between two of them compares them
- refreshing the target's statistics after a load: an SAP system's application data is read, never written: it is a pair's source (engine: hetero, source_engine: sapodp), and a hop between two of them compares them
- snapshotting the target so a cutover can be rolled back: an SAP system's application data is read, never written: it is a pair's source (engine: hetero, source_engine: sapodp), and a hop between two of them compares them
SaaS applications, read by a connector program
saas, also written program - a source, read and never written.
It does: comparing the schema; comparing row counts; comparing the data itself.
Does not apply:
- the deep checks: a SaaS application holds no state of a load to look into: it is read through its own interface
- carrying sequences and auto-increment values: a SaaS application is read, never written: it is a pair's source (engine: hetero, source_engine: saas), and a hop between two of them compares them
- comparing server settings: a SaaS application is read, never written: it is a pair's source (engine: hetero, source_engine: saas), and a hop between two of them compares them
- moving a whole database in bulk: a SaaS application is read, never written: it is a pair's source (engine: hetero, source_engine: saas), and a hop between two of them compares them
- copying table by table, resumably: a SaaS application is read, never written: it is a pair's source (engine: hetero, source_engine: saas), and a hop between two of them compares them
- keeping the target following the source: a SaaS application is followed as a pair's source, its changes carried into another engine (engine: hetero, source_engine: saas); two of them are compared, never one written
- proving the target has caught up before cutover: a SaaS application is followed as a pair's source, its changes carried into another engine (engine: hetero, source_engine: saas); two of them are compared, never one written
- telling a difference still arriving from one that is wrong: a SaaS application is followed as a pair's source, its changes carried into another engine (engine: hetero, source_engine: saas); two of them are compared, never one written
- verifying only what changed: a SaaS application is followed as a pair's source, its changes carried into another engine (engine: hetero, source_engine: saas); two of them are compared, never one written
- carrying users and their grants: who may use an application is the application's own, not carried
- noticing a move that moved nothing: a SaaS application is read, never written: it is a pair's source (engine: hetero, source_engine: saas), and a hop between two of them compares them
- refreshing the target's statistics after a load: a SaaS application is read, never written: it is a pair's source (engine: hetero, source_engine: saas), and a hop between two of them compares them
- snapshotting the target so a cutover can be rolled back: a SaaS application is read, never written: it is a pair's source (engine: hetero, source_engine: saas), and a hop between two of them compares them
Any pair of two engines
hetero - the pair: a source of one engine, a target of another.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; carrying sequences and auto-increment values; comparing server settings; moving a whole database in bulk; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; carrying users and their grants; noticing a move that moved nothing; refreshing the target's statistics after a load; snapshotting the target so a cutover can be rolled back.
Any other SQL engine
generic - either side, compared by the tables the hop lists.
It does: comparing the schema; comparing row counts; comparing the data itself; the deep checks; copying table by table, resumably; keeping the target following the source; proving the target has caught up before cutover; telling a difference still arriving from one that is wrong; verifying only what changed; noticing a move that moved nothing.
Does not apply:
- carrying sequences and auto-increment values: the generic engine reaches a server only through a standard driver, which has no statement for a sequence or auto-increment counter; a server whose one migkit reads is a hop of its own engine
- comparing server settings: the generic engine reaches a server only through a standard driver, which has no statement for a server setting; a server whose one migkit reads is a hop of its own engine
- moving a whole database in bulk: a whole database moved in bulk is the server's own dump and load, which a standard driver does not reach; table by table is table-copy
- carrying users and their grants: the generic engine reaches a server only through a standard driver, which has no statement for an account or its grants; a server whose one migkit reads is a hop of its own engine
- refreshing the target's statistics after a load: the generic engine reaches a server only through a standard driver, which has no statement for a refresh of the optimiser's statistics; a server whose one migkit reads is a hop of its own engine
- snapshotting the target so a cutover can be rolled back: the generic engine reaches a server only through a standard driver, which has no statement for a snapshot of the server; a server whose one migkit reads is a hop of its own engine