Amazon Managed Service for Prometheus now supports out of order sample ingestion

Amazon Web Services has enhanced its Managed Service for Prometheus with two new capabilities designed to improve data reliability and alerting accuracy in distributed monitoring environments. The service now supports out-of-order sample ingestion with a configurable time window that defaults to one minute, allowing workspaces to accept metric samples that arrive outside strict chronological order rather than discarding them. Additionally, AWS introduced a workspace-level rule query offset feature that delays rule evaluation queries to accommodate late-arriving data. These features specifically address common challenges in distributed systems where metrics collection involves multiple collectors, batched data exports, or variable network latency. The out-of-order ingestion prevents data loss by preserving late-arriving metric points, while the rule query offset ensures that alerting rules have complete data sets before executing evaluations. AWS has also added two CloudWatch metrics—OutOfOrderIngestionRate and OutOfOrderSampleAge—to help administrators monitor ingestion patterns and optimize their configurations. The new capabilities are immediately available across all AWS regions where Amazon Managed Service for Prometheus operates in general availability. Organizations can configure both the out-of-order time window and rule query offset through the AWS console, API, or command-line interface, with the ability to disable the feature entirely by setting the time window to zero.