AWS MSK failover mechanism and handling data loss
0
- can we have aws msk clusters as a active - active in two regions (first - read write, second - read only) using replicators
- If So, if one cluster fails how can i promote the second one as to read, write to accept read write requests
- Will the first cluster come to active again by itself
- if it becomes active how can copy the data written in second cluster and make the first again the read, write, second back to read only 5.Correct me if there are any mistake in above flow
- Topics
- Analytics
- Language
- English
asked 2 years ago391 views
1 Answer
- Newest
- Most votes
- Most comments
0
Hi!
- I believe the setup you are mentioning would fall under an "active - passive" configuration: and you can do this using Kafka Replicators (MirrorMaker 2).
- You can either manually update the read only to accept write requests in the event of a failover or setup automation with an architecture that would involve something like Lambda and Cloudwatch.
- When the primary cluster comes back online, it does not automatically resume its primary role. This will again either require manual intervention or similar automation as mentioned in Step 2.
- You will also need to reverse replicate the data when it is back online. Again, something like Mirrormaker 2 will take care of this for you if you setup bi-directional replication!
Your overall understanding is correct, but this is not an "active-active" setup. Let me know if you need further help!
answered 2 years ago
Relevant content
asked 5 years ago
asked 2 years ago
Thanks, but it is not exactly a active passive, because im serving the read request in the second region. So when ever primary fails i want to forward traffic to secondary with write cabability and when primary is back i want to copy the additionally written data to primary and then froward the primary region traffic to primary. meanwhile scondary will continue serving for seconadry region read requests