Elastic has launched new observability features for metrics monitoring, adding native Prometheus support and a rebuilt metrics engine.
The release targets a growing cost problem for companies running cloud-native and AI-heavy infrastructure, where large volumes of high-cardinality metrics data can drive up observability bills.
Engineering teams have long relied on metrics systems built for more static environments. But Kubernetes, microservices and AI workloads have sharply increased the number of time series that must be stored and queried. That shift has made the economics of observability more pressing for DevOps and site reliability engineering teams, especially as organisations try to retain detailed data without losing visibility.
The new release brings metrics and logs together on a single platform instead of splitting them across separate backends and query tools. It also adds native PromQL support in Kibana and support for Prometheus Remote Write, allowing existing dashboards, alert rules and scrape configurations to continue working without modification.
Elastic says the platform is built on Elasticsearch's columnar metrics engine, which can query metrics up to 30 times faster than Prometheus and store data up to 2.5 times more efficiently. In technical material published alongside the launch, the company says the system uses 3.75 bytes of storage per data point.
Cost pressure
Elastic is positioning the release around a familiar frustration for infrastructure teams: the trade-off between keeping full-resolution operational data and limiting spend. In many environments, high-cardinality data from containers, services and AI workloads has led teams to sample or aggregate metrics to avoid steep cost increases.
According to Elastic, its approach removes cardinality limits and avoids custom metric penalties common in parts of the observability market. It is also offering automated migration tools for users moving from Datadog and Grafana, converting dashboards, alert rules and PromQL queries into Kibana equivalents.
The product changes extend beyond storage and query speed. Elastic has added pre-built Kubernetes workflows, alert templates, service level objective content and machine learning anomaly detection jobs that activate at ingest, aiming to shorten the time from alert to diagnosis.
It has also introduced what it calls agentic investigations, designed to correlate metrics, logs and traces in a shared backend when an alert fires. These workflows surface what changed and indicate the severity of a deviation before engineers are paged, the company says.
The feature set reflects a broader effort by observability vendors to apply AI tools to incident investigation, especially as system architectures become more distributed and telemetry volumes rise.
"Elastic was already the platform many SREs trusted for logs at scale. Now we're bringing that same impressive scale, performance, and operational simplicity to metrics, delivering up to 30x faster metric queries than Prometheus, native Prometheus compatibility, and a more predictable cost model for high-cardinality metrics," said Baha Azarmi, General Manager, Observability, Elastic.
"With a single backend for every signal, a single query language, and investigations that start before anyone is paged, SREs get complete context at the moment they need it most - without the bills that have forced teams to compromise on the data they keep," Azarmi added.
Customer response
Elastic included early customer accounts from companies with large-scale cloud and Kubernetes estates. Those users highlighted the challenge of handling fast-growing data volumes while maintaining visibility across increasingly complex environments.
"As we've moved more applications into Kubernetes and expanded our cloud footprint, data is growing rapidly and our need for granular, high-cardinality metrics is increasing," said Jeff Beagley, Manager of DevOps, SRE, and Cloud Engineering, Bass Pro Shops.
"Elastic's new metrics capabilities let us handle that volume and surface the insights we need. Coupled with Elastic's OpenTelemetry support, we get visibility into an increasingly complex architecture - all while keeping performance up and costs down," Beagley said.
Eurowings Aviation described similar gains from unifying telemetry signals in one system rather than moving between separate tools.
"At Eurowings, the improved metrics performance, native Prometheus support, logsdb and incident-handling workflows in Elasticsearch have helped our teams achieve faster incident response times and a more unified view across signals without jumping between systems," said Iosif Tournas, Cyber Security & Elastic Platform Lead, Eurowings Aviation.
"These new metrics capabilities complement the millions of log events per minute and APM traces we're already handling in Elastic Observability. This unified view reduces operational friction, breaks the silos between teams and the time it takes to detect and respond to issues," Tournas said.
Elastic says the columnar metrics engine, ES|QL time series support, PromQL in Kibana and Prometheus Remote Write ingest are generally available, while its Observability MCP App, Agent Skills and migration platform remain in technical preview.
The tools run across Elastic Cloud, serverless and self-managed deployments, according to the company.