Can Elasticsearch and OpenSearch clusters be scheduled to stop outside business hours?

For EC2-based deployments, you can schedule the underlying Auto Scaling Groups to scale down to zero and bring them back up later. EventBridge Scheduler can trigger Lambda functions that stop and start the services, update the desired capacity, and handle any required health checks. Make sure the cluster data is stored on persistent volumes and allow enough time for the nodes and cluster to become healthy again.

Another option is AWS’s Instance Scheduler solution, which can start and stop EC2 instances according to tags and schedules. It could cover the predictable off-hours window, while a booking portal could invoke a controlled start workflow for on-demand access.

Cloud Custodian is also worth considering if you want policy-based automation. You can define off-hours rules using resource tags and have it stop or scale resources down, then combine that with an authenticated API or Lambda-backed workflow for user-initiated starts. Be careful not to let users directly change infrastructure settings without validation and permissions.

If you’re using OpenSearch Serverless, the scale-to-zero capability may remove the need to manage instances directly. It can reduce costs during idle periods, although there is a warm-up delay when traffic resumes, so it may not be suitable for workloads that need immediate availability.