AWS has added time-series collections to the next generation of Amazon OpenSearch Serverless, combining a lower-cost warm storage tier with built-in alerting and anomaly detection. The practical gain is that teams can keep older observability data searchable without paying hot-storage prices for all of it.
AWS AI Watch analysis
What happened
AWS says the collections can ingest logs, metrics, traces and event data without customers provisioning or managing clusters. Recent data stays in hot storage; customers can choose when it moves to the new warm tier. Collections support up to 100 TB of hot storage, and the service is available now in every AWS Region where OpenSearch Serverless is offered.
The launch also brings alerting and anomaly detection, so teams can be notified of issues and spot unusual patterns in logs and event data without separate tooling. AWS’s announcement points to pricing information for the new warm storage tier.
Why it matters
Observability data tends to pile up, while its value does not politely expire when a billing period ends. Separating recent, fast-to-query data from older, lower-cost searchable history gives teams another way to manage that trade-off. Built-in anomaly detection and alerting could also make it simpler to spot trouble without adding another service to the stack.
That is useful infrastructure news, though it is not a promise that every anomaly will be caught or every query will be cheap. The actual economics depend on the storage mix and pricing, details AWS directs customers to check.
Our read
The appealing part is the combination: searchable long-term history, a sizeable hot tier and alerting in the same collection. That could simplify observability setups for teams already using OpenSearch. Before moving a workload, compare the warm-storage pricing with your retention and query patterns; “lower-cost” is a direction, not a bill estimate.
What to watch
- The cost of warm storage and the conditions attached to the new pricing.
- How teams choose hot-to-warm transition periods for their data.
- Whether built-in anomaly detection proves useful on real operational workloads.
Discussion spark: Would you keep months or years of observability data searchable in a lower-cost warm tier, or is cheaper storage not worth it if older data is queried less often?
Sources and evidence
- Introducing time series collections for the next generation of Amazon OpenSearch Serverless (28 October 2026, 18:35 UTC)
not affiliated with or endorsed by Amazon Web Services (AWS)