Rolling Updates: Database Version Migrations with minimum Disruption
summary
YDB is an Apache-2.0 distributed SQL that runs across thousands of nodes and keeps read/write availability during upgrades by combining a rolling restart tool with a cluster management API and a tablet-based architecture. Storage uses nine virtual disks per group and writes three copies (escalating to four on failure) so AZ outages remain writable. Compute nodes serve single databases, overprovision ~40% for DC loss, and implement tablet evacuation plus an SDK shutdown header so restarts cause brief, controlled retries instead of wide outages.
AI-written summary · about
chapters (11)
- 0:00Introduction and speaker intro
- 2:14What is YDB and its architecture
- 6:28Why rolling updates are necessary
- 8:39Challenges of rolling updates
- 9:47YDB’s three-layer architecture
- 12:41Surviving data center outages
- 20:25Restarting compute nodes and errors
- 21:08Solving rolling update challenges
- 26:30Evacuating tablets and sessions
- 31:34Future of YDB and features
- 33:41Closing remarks and Q&A
related talks
01h 00mParallel Asynchronous Replication b/t YDB Database Instances
29mEnterprise Scale: How a Distributed Architecture Can Deliver Scale and Resilience
09mAchieving Zero Downtime with Little Effort
29mData Modeling for Distributed SQL
24mFailure is Not an Option: Highly Available Distributed SQL
01h 27mYugabyteDB YSQL Development
59mServerless nature of
22mYugabyteDB: Discover What's New