Architecting for Performance, Scalability and Real-time Data with AWS
summary
Benchmark on a 20-node i3en24xl simulates a 1 PB user-profile DB plus 1.5 TB campaign DB and hits 5.2M TPS <1ms (reads and writes), dropping to ~4M TPS at an 80/20 read/write split; equivalent Cassandra sizing needs ~514 nodes, yielding a claimed ~73% infrastructure cost reduction. Architecture uses a hot cache / cold store pattern with XDR or stretch clusters, three-AZ deployments, VPC separation for management and clients. Instance notes: i3en ~1.38M TPS, i4i ~1.78M TPS, Graviton prelim ~1.2M TPS at roughly one‑quarter the cost.
AI-written summary · about
chapters (9)
- 0:00Introduction
- 1:30What are your customers using
- 3:00The Trade Desk
- 6:50Fraud Opportunity
- 9:00Nustar
- 10:40Best Practices
- 13:35Benchmark
- 15:45Graviton
- 17:15Performance Metrics
related talks
24mCloud Workloads At Petabyte Scale
15mAWS On Air Voice of the Customer ft. Aerospike, Inc.
49mBuilding real-time applications to utilize AWS Graviton
48mAnalytics Architectures on Microsoft Fabric
26mReimagine Real-time
35mReplacing Cassandra: A Digital Transformation for the World’s Largest Digital Identity Network
30mBuild for Scale, Performance With Feature-packed Amazon DynamoDB
52mHow Netflix Provisions Optimal Cloud Deployments of Cassandra