Google has launched a SQL database porting service it said will ease the lift and shift of SQL family databases into its managed relational database service Cloud SQL.
Dubbed the Database Migration Service, or DMS, the product is designed to help users get their instances of open-source MySQL and PostgreSQL, and proprietary Microsoft SQL Server, from an on-premises instance to its managed service in Google Cloud Platform. It can also be used to port these databases from another cloud provider to the same service.
DMS uses log-shipping to replicate data from the source database to the destination. It streams the initial snapshot of data, then catches up and continuously replicates new data as it arrives in the source, an approach Google claims reduces latency.
The search and cloud biz said that the source and destination are continuously up to date “because they rely on the databases’ own native replication capabilities.”
Users can then point the application at the Cloud SQL as the primary database, with minimal downtime, Google claimed.
Because DMS is serverless, users need not concern themselves with provisioning, managing, or monitoring migration-specific resources. “The source database’s data, schema, and additional database features are replicated to the Cloud SQL destination reliably, and at scale, with no user intervention required,” Google said.
In porting the SQL databases to the managed service, users can define how a new instance will connect to a source database for continuous data replication. Options include the IP address of the cloud SQL destination on the database source's machine, private connectivity through a reverse SSH tunnel on a virtual machine, or via a VPC peering connection, said Gabe Weiss, developer advocate for Google Cloud Platform.
Although the service may have its uses, it is not really a database migration as such, said Philip Howard, research director at Bloor Research. "Migration is the movement of persistent data that involves some sort of restructuring of that data while the usage of that data remains constant," he said.
As such, any inefficiencies that have built up in the database over time would persist in the new instance. “It would appear to be just a way to move what you've currently got to somewhere else. That doesn't mean to say it isn't valuable if you want to port from one environment to another,” Howard said.
Alkin Tezuysal, senior technical manager at PlanetScale, said users should ask about the cost implications before diving in with the service.
He said: “Cost is… truly unknown until after the migration is done. Is there a cost estimator that they could offer for this serverless service? Is it based on the data, IOPS, network or CPU power?”
Google has yet to respond to The Register’s request for answers to these questions. ®