Database Lab

Verify complex database changes

• With Database Lab, all database schema or data changes are automatically verified covering 100% of cases. Engineers can easily see the true behavior of any database migration before it is deployed and optimize it when needed. Management is confident that the risks of downtime and performance degradation are eliminated.

• Instant provisioning of full-size independent clones of large databases.
• Production is not affected nor does it need any special changes to make Database Lab work. Special technologies (such as ZFS) are used only on Database Lab servers.
• Automated recycling of used clones.
• Multiple parallel verification processes.
• Useful artifacts highlighting potentially dangerous issues (including: risks of locking issues and contention, query analysis, DBA-level metrics such as checkpoint and autovacuum behavior analysis).

Optimize SQL queries

• Development and testing using real-life data and SQL plans identical to production.
• Two major bottlenecks are eliminated: no more long waiting for your more experienced colleague's attention and no more long waiting for a new database clone provision.
• SQL optimization recommendations.
• Searchable history of all SQL optimization sessions to collect and share the optimization experience.
• SQL plan details are explained and visualized.
• Hypothetical indexes to get a sneak peek of the forecasted SQL query plan before building large indexes.

Develop using thin local clones to iterate more and faster

• Develop and test using full-size database clones provisioned in seconds to get most reliable results much faster.
• Save disk space. A lot. Local thin clones share the majority of data blocks ("copy-on-write") saving time and budgets.
• Tasks such as verification of major upgrades, partitioning, index maintenance can and must be verified in Database Lab first, to minimize risks of negative events in production.

Recover accidentally deleted data

• In the case of manually deleted data that needs to be restored ASAP, it can be done almost instantly (proper snapshot management configuration is required in Database Lab in advance).
• Using thin cloning, the point-in-time recovery (PITR) can be performed without long waiting.
• Typically, for 1 TiB database, PITR requires more than 1 hour in classic setups. With Database Lab, it takes a few seconds to a few minutes.

Better performance for analytics

• Run heavy analytical SQL, perform data export without affecting the production servers.
• Analysts work with thin clones, which are fully independent.
• When a long-lasting query needs to be executed, an analyst can work independently, not interfering with production workload or a colleague's work.
• Production servers are not in danger: autovacuum activity is not affected, long-running queries are not causing bloat.