Right Size ElasticSearch Clusters
AWS Resource Type
OpenSearch cluster (aka OpenSearch Service Domain)
OpenSearch and ElasticSearch clusters are difficult to size. Therefore they are often overprovisioned. Cloudfix identifies these overprovisioned clusters and resizes them by targeting them at a 70% peak utilization.
Criteria for identifying the opportunity
- CloudFix rightsizes OpenSearch clusters by targeting 70% peak utilization of both CPU and memory, based on CloudWatch data from the last 30 days.
- It excludes clusters using ultrawarm/cold storage or clusters that are not in a green state
Potential savings (range in % on annual basis)
CloudFix can reduce CPU and memory resources of overprovisioned OpenSearch clusters to realize savings of about $0.12 per cluster hour for each resource tier reduction. The savings may vary depending on the extent of over-provisioning of the resized cluster. Our experience suggests that 15%-20% savings are to be expected.
What happens when the Fixer is executed?
- The fixer retrieves the target cluster attributes and verifies the Processing, UpgradeProcessing, and Deleted flags are all false.
- The fixer updates the cluster and master node Instance type to resize it based on the calculated right size.
Changing the instance size of the cluster is done by using blue/green deployment process, which minimizes chances of disruption to the service.
Is it possible to rollback once CloudFix implements the fixer?
Yes, the rollback is automatic based on CloudWatch alarms on the CPU and memory usage of the cluster.
Can CloudFix implement the fix automatically once I accept the recommendation?
Does this fix require downtime?
No. AWS uses blue/green deployment process.