Day 1 Session 1
summary
Explains how CAP and PACELC force tradeoffs between availability latency and consistency and maps those to Cosmos DB features: 10ms read SLA claim for co-located replicas, demo shows 72ms cross-US reads vs ~2ms from a local replica and single-master writes ~73ms vs multi-master writes ~6ms. Describes five consistency levels (strong to eventual), dynamic global quorums for strong, RPO/RTO behavior, autoscale (Autopilot) and a free 400 RU / 5GB tier for POCs.
AI-written summary · about
chapters (17)
- 0:00Introduction and overview
- 1:02Why build distributed applications
- 2:32Trade-offs in distributed systems
- 3:38CAP theorem explained
- 6:31PACLC theorem and latency trade-offs
- 8:37Demo: Reducing read latency
- 14:35Demo: Multi-region writes
- 17:33Consistency levels overview
- 23:37Latency vs consistency trade-offs
- 27:32Demo: Consistency vs latency
- 31:29Building highly available apps
- 34:14RPO and RTO explained
- 40:00Additional resources and Q&A intro
- 41:24Q&A: Slides and multiple API support
- 43:55Q&A: Cost and performance tips
- 49:57Q&A: Auto-scaling and consumption plans
- 53:51Summary and closing
related talks
01h 15mCAP Theorem — two decades and few clouds later
01h 02mAzure Cosmos DB - Low Latency and High Availability at Planet Scale
47mDatabase Resiliency
52mCAP Theorem - two decades and few clouds later
03h 02mAPAC Live Stream
20mCAP Theorem Reloaded
45mGetting the Most from NoSQL in the Cloud with Azure Cosmos DB
22mCloud-Native Apps, recovery ready!