Closed navneet1v closed 1 month ago
@rishabh6788 what are your thoughts on this item?
Agree, having disk based instances would benefit such use-cases. I can add this to our sprint, but in the meantime if you can squeeze in a PR I can review it and merge quickly.
I can give it a try, but might take some time for me to get it work. Will see if I can pick it up in upcoming week.
The feature is available to use now.
Is your feature request related to a problem?
Currently in the CDK we only support EBS based instances with the exception of i3 instances. This limit the testing to use only EBS based cluster. With launch of Disk based vector search, we have create our own custom solutions to setup the cluster. But this repo provides a very easy way to create the cluster, hence having a capability to use SSD based instances via this CDK will be very useful.
What solution would you like?
Ideally, we support instance with SSD. Only by adding the nodes here: https://github.com/opensearch-project/opensearch-cluster-cdk/blob/main/lib/opensearch-config/node-config.ts#L63 won't help, we also need to mount the SSD to the correct data folder where opensearch stores the index. Ref: https://docs.aws.amazon.com/AWSEC2/latest/UserGuide/making-instance-stores-available-on-your-instances.html
What alternatives have you considered?
Alternative is to create the setup from scratch manually.
Do you have any additional context?
NA