Bringing together the Apache Cassandra experts from the community and DataStax.

Want to learn? Have a question? Want to share your expertise? You are in the right place!

Not sure where to begin? Getting Started

 

question

phofegger_148429 avatar image
phofegger_148429 asked ·

Does it matter which source DC is used with nodetool rebuild when adding more DCs?

[FOLLOW UP TO QUESTION #3042]

I continue the migration now.

Step 1: is done

Step2: is done

Step 3: I prepare the 5 nodes (Cassandra installation) and create the 3rd DC.

I have question.

If I add the 3rd DC. From which DC should I rebuild or it doesn't matter from which one. At the end the data on all 3 DC should be in sync. I would suggest from DC1 (all clients write into this cluster) . What do you think .

Many thanks in advance

cheers

Patrick

cassandra
10 |1000 characters needed characters left characters exceeded
Toggle Comment visibility. Current Visibility: Viewable by all users

Up to 8 attachments (including images) can be used with a maximum of 1.0 MiB each and 10.0 MiB total.

1 Answer

Erick Ramirez avatar image
Erick Ramirez answered ·

@phofegger_148429 Good to hear that your DC migration is finally happening. Technically, it will work whichever DC you choose as the "source" for streaming. However, there are some things you need to consider.

Since DC1 is the primary data centre where writes are going to, I am going to assume that the application is also actively reading from this cluster. If this is the case then I would recommend you use DC2 as the streaming source to build the 3rd DC so all the transactional traffic from your application is not affected by the additional reads.

I do understand that DC1 has 10 nodes and DC3 will be reduced to 5 nodes so DC3 has less capacity so it might take longer to build across the WAN at the new data centre. But at least you know that the application won't be impacted if you don't stream from DC1.

Let me know what you think. Cheers!

2 comments Share
10 |1000 characters needed characters left characters exceeded
Toggle Comment visibility. Current Visibility: Viewable by all users

Up to 8 attachments (including images) can be used with a maximum of 1.0 MiB each and 10.0 MiB total.

phofegger_148429 avatar image phofegger_148429 ·

Thanks for the answer. Yes DC1 (10 nodes) are the primary DC. DC2 (5 nodes) is the analytics Cluster (co-located with mesos and spark). DC3 becomes the new analytics cluster and ist located in the same physical DC with DC2. After DC3 migration finished then I will decommission DC2 and rebuilt as DC4 in a new physical DC and DC4 become the primary DC. Then DC1 will be decommissioned. At the end, I have DC3 (5 nodes for analytics) and DC4 (5 nodes - Primary DC) .


0 Likes 0 ·
Erick Ramirez avatar image Erick Ramirez ♦♦ phofegger_148429 ·

Sorry, I've just managed to circle back to this. I think you're approach is fine and hopefully it's all going as planned. Cheers!

0 Likes 0 ·