Lightning Talk: Multi-Datacenter Strategies for Data Services
summary
Classifies multi-DC setups as close-range (<50–100 km) versus wide multi-region and warns that stretch infrastructure only works when links are very reliable. Shows automation using BOSH and service brokers to provision clusters, cites synchronous Postgres replication and Cassandra replication as concrete options, and highlights quorum math (2n+1 quorum, three nodes minimum). Claims there is no generic multi-DC recipe, reports replication lag and split‑brain as real failure modes, and treats platform, data service and app design as equally affected.
AI-written summary · about
chapters (12)
- 0:00Introduction to talk
- 1:50Defining multi-datacenter goals
- 3:37Infrastructure migration lessons
- 5:25Close-range datacenter scenarios
- 7:12Stretched infrastructure challenges
- 9:38Three-datacenter setup
- 11:22Multi-region datacenter strategies
- 13:56Primary-secondary replication
- 15:40Generic multi-region capability
- 16:43Application vs data service layer
- 17:47Async replication lag risks
- 19:15Designing for multi-datacenter awareness
related talks
10m⚡ Lightning Talk: Navigating the Complexities of Stateful Workloads in Kubernetes
38mDeploying Database Clusters in the Cloud
40mMulti Cloud Magic: Leveraging Multi Cloud Clusters in Real World Scenarios
43mMulti-Cloud Magic: Leveraging Multi-Cloud Clusters in Real-World Scenarios
57mDSS Keynote — Back to the Future: Prepare Your Data Infrastructure for Anything
28mBootstrapping a Multi-DC Cloud Native Observability Stack
29mMulti-DC Cloud Foundry: What, Why and How?
23mState in a stateless world: data serving in the cloud