Journey from PostgreSQL to CockroachDB: Scaling Cisco AI Defense
summary
Cisco migrates from PostgreSQL to CockroachDB to solve transaction ID/vacuum storms, sharding pain and constant failovers caused by running Postgres in Kubernetes; the platform now stores LLM prompts and responses for enterprise AI across 100,000 tenants and replaces per-tenant schemas with multi-tenant tables to avoid millions of small tables and autovacuum collapse. CockroachDB runs across six regions, provides built-in recovery and horizontal scaling, and feeds downstream training via CDC to Snowflake and Databricks, cutting ops toil and eliminating frequent outages.
AI-written summary · about
chapters (6)
- 0:00Introduction and background
- 0:30Early days in Kubernetes
- 3:53Scaling and architectural shift
- 7:26The road to CockroachDB
- 13:07Modern Cisco AI implementation
- 14:58Conclusion
related talks
27mBeyond the Roadmap and Into the Future of CockroachDB
24mNavigating Distributed Databases: Challenges & Opportunities
48mBuilding mission-critical applications with CockroachDB
29mHow City Storage Systems designed its Data Infrastructure to minimize toil
18mFrom Today to 10X: Why Groww chose
15mWhy Yubi selected CockroachDB to power 10x transactional growth
09mWhy Choose a Database Designed for K8s?
33mWhat We’ve Learned Building a Multi-Region DBaaS on Kubernetes