PolarDB: Alibaba Cloud-Native Relational Database Service
summary
PolarDB describes a three-tier, compute‑memory‑storage disaggregated architecture deployed on Alibaba Cloud (500,000+ cores, 80+ AZs) that adds an extra disaggregated memory layer used as a communication hub and terabyte‑scale remote buffer pool. Multi‑master scaling shows near‑linear throughput up to 16 RW nodes and a row‑level protocol called polar fusion that coordinates caches over 100G RDMA. SmartSSDs deliver inline compression >3x and claim up to ~50% TCO reduction; TPCH and sysbench results report major latency and throughput gains.
AI-written summary · about
chapters (18)
- 0:00Introduction
- 0:28Alibaba Cloud
- 0:59PolarDB
- 1:58Pluggable pieces
- 2:50Tiered architecture
- 3:43Why use a threetier architecture
- 5:09Scaling up
- 6:22Memory disaggregation
- 7:48Hardware integration
- 10:06Types of capabilities
- 11:57Performance
- 12:31Scaling out
- 14:20Multimaster performance
- 15:15Hypothetical solution
- 16:53Power TV Serverless
- 17:59Data Cost
- 19:40Polar for AI
- 20:23PolarDB PostgreSQL
related talks
42mPOLARDB architecture
38mPOLARDB Meets Computational Storage: Efficiently Support Analytical Workloads in Cloud
PolarStore: High-Performance Data Compression for Large-Scale Cloud-Native Databases
28mIntroducing AlloyDB, a PostgreSQL-compatible cloud database service
30mLarge-Scale Practice of Persistent Memory in Alibaba Cloud
36mArcticDB, the OLAP antidote
36mData Warehouse, Data Lakes, What's Next?
17mBreaking Through AI Workload Bottlenecks How OceanBase’s AI Native Capabilities Empower Enterprise