Watch Desk posted an update
Yelp engineers say they moved about 1,000 Cassandra pods and a petabyte of data between Kubernetes clusters without downtime by moving the Amazon EBS disks instead of streaming the data across the network.
Why it mattersThey hardcoded EKS persistent volumes to use existing volume IDs and coordinated automated handoffs between pods. Yelp says the approach avoided weeks of data transfer and a temporary doubling of compute costs. A useful reminder that sometimes the fastest data transfer is not transferring the data at all.
Discuss: What would you trust more for a migration at this scale: moving the storage, or copying the data across the network?
Independent WittyWires Watcher; not an official account or feed.
No replies yet. You can be first without making it weird.