Journey with Aerospike: Architecture Best Practices
summary
A deployed Aerospike fleet grows to 171 GB across 139 nodes serving about 18 billion reads and writes per day with peaks of 17 million queries per minute and 99.95% reads under 1 ms. Techniques shown include using expect_gen equality for record-level optimistic concurrency, snapshot-and-replay for multi-record consistency, a library that shards large bins (real cases up to ~150 MB), denormalized schemas using native lists/maps plus server-side expressions, zone-isolated clusters behind a load balancer, conservative XDR use, long‑lived client connections, and retry policies to reach stable 10,000 writes/sec.
AI-written summary · about
chapters (7)
- 0:00Introduction and overview
- 1:32Aerospike performance scale
- 4:04Handling transactions
- 8:59Overcoming record size limits
- 14:51Data modeling best practices
- 31:20Data isolation and availability
- 39:50Developer best practices
related talks
22mThe Art of the Possible
40mDeveloping highly scalable applications with Aerospike
33mInnovation with Aerospike: How AppLovin Handles Billions of Requests per Day
29mWhat breaks when Aerospike hits 6 million QPS
18mProduct Review and Roadmap
05mHighlights From Aerospike APJ Summit 2023
31mAerospike's Innovation Agenda
50mAerospike Technical Overview