Zero-Downtime Patching Part 1: Prewarming

Hans Norheim, a Senior Staff Software Engineer at Databricks, discusses how Neon and Databricks Lakebase are utilizing the lakebase architecture to achieve zero-downtime patching by employing a technique called prewarming. This process aims to make database version updates and security patches seamless and unnoticeable to users by preventing performance degradation after restarts, which typically suffer from cold caches. The lakebase architecture leverages stateless compute nodes and a disaggregated storage system, allowing compute replicas to be spun up and prewarmed without affecting the active primary database. By preloading data into cache and subscribing to the write-ahead log from shared storage, the new compute node is ready to take over as the primary without restarting the database server. The prewarming approach significantly reduces the time needed for cache recovery, maintaining consistent performance during maintenance and eliminating the need for full replicas or complex orchestration. This innovation ensures that the planned maintenance windows can be executed without disrupting the availability and performance of customer databases.

Use this post, company, and trend context to find content marketing opportunities, perform competitive analysis, or address product feature gaps via the Plushcap MCP server or the Plushcap API.