Disaster Recovery with a Distributed Database
summary
ArangoDB implements asynchronous DC-to-DC replication by dumping change events into Kafka on each side and replaying them with ArangoSync so a primary cluster in Paris can catch up a standby in Vilnius with small, acceptable lag; replication is purposely asynchronous to avoid write latency. Mesos orchestrates the deployment, resource reservations and restart logic, originally via a ~5,000‑line C++ scheduler framework. Demo shows scaling DB servers up and down, data rebalancing and successful cross‑datacenter sync with encrypted links. ArangoDB uses Kafka for async DC-to-DC replication Mesos manages deployment and resources
AI-written summary · about
chapters (14)
- 0:00Introduction
- 1:25Distributed Data
- 2:21Different Data Models
- 3:14MultiModel Databases
- 5:31Native MultiModel Databases
- 6:19Why Different Models
- 8:04Native Approach
- 14:23Replication
- 14:46Options
- 17:18Kafka
- 19:12Arango Sync
- 21:36Replication between two data centers
- 23:56Demo
- 32:11Questions
related talks
40mDistributed Data on Distributed Infrastructure
29mMulti Active-Active and disaster recovery with MongoDB database center
43mCreating Fault Tolerant Services on Mesos
39mHandling Billions of Edges in a Graph Database
09mMulti-Region Replication & DR for Distributed Postgres in Azure Cosmos DB
32mAutomating high availability MongoDB cluster deployment and management
29mEnterprise Scale: How a Distributed Architecture Can Deliver Scale and Resilience
27mBuilding Event-Driven Systems with PostgreSQL Logical Replication and Drasi