How a SaaS Streaming Platform Achieved 3.26M QPS with Zero Downtime
Learn how we migrated one of the largest streaming user management platforms to TiDB's distributed SQL architecture
Key Metrics:
- 3.26M QPS achieved (vs 300k QPS on maxed-out MySQL)
- 79% storage reduction (from 7TB to optimized footprint)
- Zero-downtime migration (28 hours, 7TB of data)
- P99 latency < 20ms (maintained at massive scale)
- 4x query performance improvement
- 121 instances - dynamic scaleout capability
What You'll Learn:
- Why single-writer MySQL architecture fails at scale (replication lag, commit latency, 30-second failover)
- The exact scaling challenges faced: 25k TPS requirements, 3M QPS targets, sub-20ms latency demands
- TiDB's distributed architecture solution: multi-master, automatic sharding, zero code changes
- Zero-downtime migration strategy using DM + CDC (Data Migration + Change Data Capture)
- Real-world workarounds: Redis failure handling, write hotspot resolution, connection balancing
- TiFlash integration for real-time analytics (7-min batch reports vs 1-hour delays)
- Performance benchmarking results: commit latency, query optimization, and connection scaling
About the Speaker:
Kabilesh PR is a Founding Partner at Mydbops, which manages 10 Billion DB transactions daily across 6000 servers with a dedicated 24/7 DBA team. With 9 years of expertise and 800+ happy clients, Mydbops specializes in MySQL, MariaDB, PostgreSQL, and TiDB consulting services.
Target Audience:
- Database Administrators handling scale challenges
- Platform Engineers building distributed systems
- DevOps teams managing critical infrastructure
- Solutions Architects designing resilient systems
- CTOs and Tech Leaders planning large-scale migrations
Key Details:
- Speaker: Kabilesh PR, Founding Partner at Mydbops
- Duration: 45 mins
- Date & Time: Dec 12 | 03:30 PM - 04:15 PM
- Format: Live Online Webinar
REGISTER NOW