Google recently released a point-in-time-recovery feature for its Cloud Spanner database. The new capability aims to help protect against accidental data loss and corruption.
Cloud Spanner is a distributed, relational database service on the Google Cloud Platform (GCP). The service combines global scale with ACID compliance using physical hardware clocks found in Google’s data centers. This configuration aims to log reads and writes to all instances with a consistent timestampand deliver a consistent notion of time across all instances.
Google added backup and restore capabilities for Cloud Spanner in May 2020. The new point-in-time recovery (PITR) features seek to provide users more granular control over data recovery processes.
Users may configure a version retention period to keep versions of data and schema for Spanner instances. They may specify a version retention period from a minimum of one hour to a maximum of seven days.
In the event of unwanted data loss, users can opt to recover the complete database or to restore specific portions of the database.
The enhanced lifecycle management features for Cloud Spanner come amid increasing investment in cloud databases.
Gartner recently predicted that "by 2022, 75% of all databases will be deployed or migrated to a cloud platform, with only 5% ever considered for repatriation to on-premises."
Microsoft’s Azure Cosmos DB, a service often compared to Cloud Spanner, captures automated backups in regular intervals and at user-defined points in time. Users may request a restore of a Cosmos DB instance via a support ticket. A continuous backup feature for Cosmos DB is in public preview.
Amazon Aurora Serverless v1, a database recommended for low-traffic and testing environments, offers point-in-time recovery for instances "to any second during your retention period, up to the last five minutes."
Aurora Serverless v2, designed for business critical workloads, is in preview. According to the product’s website, it offers similar point-in-time recovery capabilities compared to Aurora Serverless v1.
PITR for Cloud Spanner is available in all Google managed regions. Users are charged for the storage consumed by hosting the versions of the data and schema during the version retention period. Additional information about PITR can be found in the documentation.