High CPU IO Wait time

Couchbase Server
1

Hello All,
We are using a Couchbase version 3.0 setup in EC2 (r5.large). We have recently migrated to AWS and we could see a very high CPU IO wait time (75%) from day1.
The couchbase cluster is having a total RAM of 85.3 GB out of which 56.8 GB is used.
Disk write queue size is ~2k to 12.5k. Compaction is happening very frequently in the buckets.
EBS volume of 50GB is attached with type gp2 and IOPS 150.

We would like to know f increasing volume to 100GB (so that IOPS also will become 300) will solve the issue. Is there any other options for solving this issue ?

2

Certainly IO Wait time is when the CPU is waiting for IO operations to complete, and if your environment is constraining that to a low value (150 tops) then that would probably explain it.

Any particular reason you’re not using AWS instance (ephemeral) storage? That’s generally a lot quicker, and given Couchbase itself will maintain multiple replicas of your data (assuming replicas are enabled) there’s little value in the network-attached EBS disks.