What's new from Grafana Labs

We’ve applied a fresh coat of paint to the Grafana homepage and added a couple of new features while we’re there.

When you operate Grafana yourself and have to troubleshoot performance issues, distributed traces are often the quickest way to see what’s going on under the hood. Until now, collecting them meant standing up an OpenTelemetry Collector and a tracing backend just to investigate a single issue.

The Azure Monitor data source now supports the Azure Monitor Metrics Batch API. Instead of sending one request per resource, Grafana groups eligible metric queries and sends them together. This means fewer calls to Azure, faster queries, and less chance of throttling on dashboards that query many resources.

See how Adaptive Telemetry is reducing your Grafana Cloud costs directly in the Cost Management and Billing App!

You can now view estimated cost savings from Adaptive Telemetry alongside your organization’s cost and usage data. The updated Usage overview brings monthly cost, contract burndown, and Adaptive Telemetry savings together, making it easier to understand both what you’re spending and the impact of your optimization efforts.

Scripted dashboards will be removed in Grafana 14. Scripted dashboards are a legacy mechanism that we deprecated years ago and are now retiring. If you use them, plan to migrate before you upgrade to Grafana 14, after which they’ll stop working entirely.

Grafana now offers dynamic dashboards and robust as-code solutions for managing dashboards at scale. Learn more about the latest dashboard capabilities and dashboards as code.

Grafana Cloud Knowledge Graph normalizes your telemetry into a fixed set of well-known labels, which means organization-specific labels like team, region, or cost_center are dropped during processing. With custom label retention, you can choose which of your own labels to keep on knowledge graph entities and recording rules.

Retained labels let you filter, group, and drill down on your own dimensions across the entity catalog, entity graph, and RCA workbench. They also propagate to the knowledge graph’s recording rules in Mimir, so you can use them in dashboards and alerting.

When a graph packs in dozens of series, understanding which ones share an anomaly means visually tracing overlapping lines. Plus, if you’re a viewer, you can’t toggle the legend or stacking to get a clearer look, because those options live in edit mode.

Make distributed profiling cost effective with Adaptive Profiles! Now generally available, Adaptive Profiles applies a cost effective baseline sample rate to your services, increasing the sampling rate to 100% for a service when behavior change is detected. This allows Adaptive Profiles to reduce your stored volume by 85-90%.

Organizing a dashboard is now faster with multi-select grouping. In edit mode, you can select multiple panels at once and instantly group them into a row or a tab, making it easy to restructure dashboards without moving panels one by one.

Session Replay in Frontend Observability records what a user actually did in your app. All clicks, scrolls, and page state is played back to you alongside the errors, logs, and traces Frontend Observability already captures. Instead of guessing what led to a bug report, you can watch the session it happened in.

You can see your Grafana Cloud k6 plan limits and current stack usage in one place. That view helps you know what your subscription allows before you run a large test or add shared resources.

We’ve recently extended support for cross-filtering, bringing support to:

• Histogram
• State timeline
• Status History

This is in addition to the previously supported visualizations, bar chart, table, and time series.

When you edit a dashboard that uses auto layout, Grafana manages panel sizes automatically. Until now, it wasn’t always clear why dragging the lower-right corner of a panel had no effect or how to change the layout.

Connecting to Oracle just got easier with Easy Connect Plus.

Setting up an Oracle data source in Grafana no longer means wrangling a tnsnames.ora file or spelling out a full connection descriptor. The Oracle plugin now supports Easy Connect Plus, a third connection method that encodes the host, port, and service name in a single string. Just enter something like host:port/service and you’re connected. The port is optional and defaults to 1521, so myhost/orclpdb1 works too.

Guided configuration steps in Kubernetes Monitoring let you set up a Cluster by describing your environment instead of tuning Alloy yourself. That means there’s no need for: