Availability and Performance Tradeoffs in Global Database Deployments
summary
Explains concrete tradeoffs: use logical replication when you need table-level filtering, zero-downtime schema rollout or bi‑directional conflict handling at the cost of tens of seconds to minutes of replica lag; use physical/storage‑aware replication to get subsecond cross‑region lag and fast DR but you replicate whole volumes and cannot easily shard subsets. Shows Amazon Aurora Global Database sending log records across regions, achieves ~200,000 writes per second, under 1s RPO and under 1 minute RTO in tested scenarios.
AI-written summary · about
chapters (15)
- 0:00Start
- 0:41Global Relational Databases
- 2:17Availability Zones
- 4:25Primary Options for Supporting Global Database Deployments Logical Replication and Physical Replication
- 4:45Replication
- 5:23Logical Replication
- 8:11Challenges and Limitations with Logical Replication
- 9:49Monitoring and Management
- 10:55Benefits and Challenges of Using Physical Replication
- 13:08Storage-Based Replication
- 13:46Database Recovery
- 14:58Amazon Aurora
- 17:34Logical and Physical Replication Capabilities in Aurora
- 20:08How Does Aurora Global Database Provide Fast Cross-Region Disaster Recovery
- 22:05Failure in Region
related talks
31mDeep dive on cross-Region DR with Amazon Aurora global databases
13mOracle Globally Distributed Database
38mHow Geo-Distributed Apps Use Global Cloud Infra for Speed and Compliance
42mPostgreSQL Distributed: Architectures & Best practices
20mMaximise database availability
33mImplementing strong cross-platform resiliency in agile environments with logical r
35mAmazon Aurora DSQL: Scalable, Distributed Relational Databases
55mAmazon Aurora: A deep-dive into what's new