Why it matters
OpenSearch clusters run around the clock, so a lower hourly price on every node adds up. The change keeps the same size and data.How nOps finds it
nOps reads the last 7 days of on-demand OpenSearch instance billing (UltraWarm is left out). m, r, c and t family instances are compared with m6g/m7g, r6g/r7g, c6g/c7g and t4g of the same size, at public list prices in the same region. The saving is the on-demand cost, scaled to a month, times the list-price difference to the cheapest target. Instances not running in the last 2 days, instances mostly covered by reserved instances or credits, and savings under $5 a month are left out.Before you start
An OpenSearch or Elasticsearch version that supports Graviton instances. Enough free storage for a blue/green deployment. Permission to update the domain (es:UpdateDomainConfig).
How to implement
- Check that the domain’s engine version supports the Graviton type.
- Change the instance type (OpenSearch console: Domains → your domain → Actions → Edit cluster configuration, or
aws opensearch update-domain-config --domain-name DOMAIN --cluster-config InstanceType=NEW_TYPE.search). Do the same for dedicated master nodes if they are listed. - OpenSearch runs a blue/green deployment. Wait until the domain is Active again before making other changes.
- Update Terraform or CloudFormation so a later deployment does not change it back.
Risks and considerations
- The blue/green deployment copies all data to new nodes, which adds load and can take hours on large domains. Run it in a quiet period.
- Reserved instances bought for the old type do not cover the Graviton type.
- To roll back, change the instance type back; this runs another blue/green deployment.
FAQs
Is there downtime?
Is there downtime?
No. OpenSearch keeps the domain available during the blue/green deployment, though it can be slower while data is copied.