Issues with Opensearch Serverless scaling out

Question

Hello,

We have been experimenting some strange behaviours in Opensearch Serverless scaling out/in. During a few days it seems that it doesn't scale down to the minimum number of OCUs and when that happens our Application behaves well, but then it's like the scale out/in algorithm changes and then it start to become very spiky in the way it scales in/out and then our application starts to have a lot of issues in production.

Please see this image that describes the issue.

![Scaling out/in](/media/postImages/original/IMBqxUY0qQS_ygoUkZ9gknpg)

Is anyone else experiencing the same problem? If so how did you solve it or do we need to move our data to a managed cluster?

It would be great if we can setup the minimum number of OCUs and that is not decided by AWS, that way we can tweak it to our needs.

Cheers

Juan

Answer

We have this exact problem as well.

It would be great if we get to control the minimum number of OCUs. The current algorithm scales the OCUs down too low, which means we have frustrated enterprise customers that can't use the app until AWS decides it needs to scale up, and there's not much we can do about it

Issues with Opensearch Serverless scaling out

Contenido relevante