Cash the Cache - How BigBasket uses Aerospike
summary
BigBasket runs Aerospike as both cache and persistent store, with 11-node i3en.6xlarge cluster holding billions of records, about 140k reads/sec and 20k writes/sec, ~1 TB disk and ~500 GB RAM. It replaces memcached/Redis since 2016, stores the smart basket persisted in Aerospike, serves geojson lookups for address serviceability, and integrates at the ORM layer for automatic TTLs. Two production incidents expose risks: CDN failover causes read floods and OOM, and instance network bandwidth hits cause packet drops; upgrade and monitoring solve them.
AI-written summary · about
chapters (11)
- 0:00Introduction to big basket
- 1:04Company scale and architecture
- 2:05Aerospike implementation
- 3:26Scaling and evolution
- 5:14Performance and challenges
- 7:17Infrastructure and operations
- 8:48Automation and orchestration
- 10:08Specific use cases
- 13:46Learning from incidents
- 17:04Feature requests and feedback
- 19:30Q&A session
related talks
29mDream11: Aerospike Beyond Cache
24mCloud Workloads At Petabyte Scale
50mAerospike @ One of India's Largest Fintech Companies
40mDeveloping highly scalable applications with Aerospike
33mInnovation with Aerospike: How AppLovin Handles Billions of Requests per Day
15mAWS On Air Voice of the Customer ft. Aerospike, Inc.
05mHighlights From Aerospike APJ Summit 2023
48mThe real-time database to build your AI future