just curious if anybody tried using Kafka/Kinesis for db2 (luw on Linux platform) replication before, and it would be great if you could point us to some reference, eg. how LOB/XML behaves, latency, admin efforts, etc.. We understand CDC could publish directly to Kafka, but the replay component sounds like the tricky part. Does IBM certify any replay tool?
Not sure if we would use it any time soon, since it's a whole new world and we don't have much experience with it. But still, Kafka/Kinesis sounds like tempting alternatives to ibm native replication solutions, especially when we don't have a stringent replication latency requirement.
btw, some benefits Kafka/Kinesis bring to the replication pipeline:
- free record of what DMLs are applied to replica, up to the limit of Kafka/Kinesis data retention policy limit, which could be helpful to identify trouble makers causing high replication latency;
- easy to scale out large number of replicas, just need spin up from recent source snapshot, and read from Kafka/Kinesis to catch up.
Rui Chen[Organization Members] @ Nov 07, 2017 - 03:11 PM (America/Eastern)