Amazon OpenSearch Service Domain
Does ZopNight manage Amazon OpenSearch Service Domain?
An OpenSearch domain bills per instance-hour for every data and master node plus per GB-month of EBS storage, with no stop operation: a domain bills 24/7 until resized or deleted. ZopNight keeps 90 days of domain metrics and recommends rightsizing, storage tiering, and Graviton instance families.
Rules that fire on Amazon OpenSearch Service Domain
At a glance
| Field | Value |
|---|---|
| Scheduling notes | discovery, metrics, cost tracking, and recommendations only. |
Amazon OpenSearch Service runs managed OpenSearch and Elasticsearch domains billed per instance-hour plus storage. Search and logging domains are provisioned for peak indexing and rarely resized, making them consistent rightsizing candidates.
Node-hours with no off switch
An OpenSearch domain bills per instance-hour for every node, whether data, dedicated master, or UltraWarm, plus per GB-month for its EBS-backed storage. There is no stop operation: a domain runs until it is reconfigured or deleted, which puts OpenSearch alongside ElastiCache as always-on by construction. Cost control here is entirely capacity control.
Sized for the worst indexing day
Search and logging domains get provisioned for a peak: the migration backfill, the traffic spike, the holiday season. Then they stay there, because resizing a live search cluster feels risky. Months later the domain idles at low CPU with a node count nobody can justify. Dedicated master nodes on small dev domains and older non-Graviton instance families add their own quiet padding to the hourly total.
Storage tiers exist here too
Hot EBS storage on data nodes is the expensive default. UltraWarm and cold storage hold older, rarely queried indices at much lower per-GB rates. This is the OpenSearch equivalent of S3 lifecycle tiering, and just as commonly left unconfigured on log-retention domains where most indices are never queried again after their first week.
Watching without a stop verb
Domains are discovered on the 6-hour cycle. Hourly CloudWatch metrics from the OpenSearch namespace run with a 90-day lookback, and per-domain cost comes from Cost Explorer or CUR 2.0. Recommendations cover node rightsizing, storage tiering, and Graviton families; there is no scheduling, because there is nothing to stop.
Domain details in the console
OpenSearch Service console, then Domains. A domain’s Cluster configuration tab lists node counts and types per tier, the complete set of hourly meters in one table.