Brett Rudenstein | Hadoop Summit 2014
summary
Demonstration shows WANdisco running a single active-active single HDFS across Northern Virginia and Oregon, using a Paxos-based quorum with global sequence numbers and block-level replication. The demo runs TeraGen/TeraSort, kills name nodes and severs the WAN link, yet jobs continue and clusters resync when the tunnel returns. Claim: continuous availability (RTO effectively zero compared with typical 16‑hour recoveries) while preserving consistency across 3,000 miles and concurrent writes.
AI-written summary · about
chapters (12)
- 0:00Introduction to WANdisco
- 1:53Customer outcomes with Hadoop
- 2:57Multi-data center Hadoop demo
- 6:25Active-active Hadoop cluster demo
- 9:53Continuous availability explained
- 12:42RTO and disaster recovery
- 14:49HA vs continuous availability
- 15:52WAN link failure demo
- 19:02Demo summary and importance
- 19:23Backup and data transfer
- 20:26WANdisco company culture
- 21:48Big tech in Hadoop space
