Real time Bidding Caching Use Case On The Google Cloud Platform
summary
Real-time bidding system runs on Google Cloud with Kubernetes autoscaling, a custom binary JSON serializer and RPC library, and Aerospike as primary store and cache. It handles about 100 million bid requests per minute with p95 end-to-end latency around 100 ms, storing 2 billion user profiles and 50 million household profiles (~100 KB, 90‑day TTL). A US Aerospike cluster of 20 nodes (15 TB disk, 300 GB RAM) delivers ~350k QPS, supports daily scrubs and migrations, and helps cut infrastructure costs by roughly 65%.
AI-written summary · about
chapters (10)
- 0:00Introduction
- 1:54About VDXtv
- 3:09Evolution of RTB
- 7:44Why RTB
- 11:26Engineering Challenges
- 14:00Architecture
- 20:48Aerospike Cluster
- 23:49Roadmap
- 25:14Team
- 25:27Questions
related talks
09mInnovating Together for Real-time Impact with Google Cloud
26mPowering Real-time Decisions With a Reduced Footprint
01h 02mAchieving High Load in Advertising Technology
25mCash the Cache - How BigBasket uses Aerospike
14mMaximizing Aerospike Administration in GCP
03mInterview
28mWhen Five Nines Is Not Enough
15mAWS On Air Voice of the Customer ft. Aerospike, Inc.