Grafana 8.0 integrates with Prometheus alerting

the-eye.jpg

At its annual GrafanaCONline virtual conference this week, Grafana Labs is releasing version 8.0 of the Grafana open source project, and concurrently, releasing the next version of its enterprise cloud offering. The highlight is a long-awaited unification of Grafana and Prometheus alerting.

Although two different open source projects, Grafana and Prometheus have long been joined at the hip. With Grafana as the visualization piece and Prometheus as the underlying time-series database, both have been typically used in tandem, forming a combo that competes with InfluxDB and Elastic, each of which also offers both pieces of the puzzle.

For Grafana and Prometheus users, the catch until now was that each had different systems for setting alerts. Those rules encompassed, not only selecting the data point to be spotlighted, but also the thresholds, who should receive the alerts, and filters for blocking nuisance alerts. Typically, it dictated dual workflows in each engine to get alerts generated in Prometheus and visualized in Grafana, and confusion if the alert rules were not properly synced. With Grafana 8.0, it has integrated with the Prometheus alerting rules engine, so the workflow need not be duplicated. In 8.0, alerts from Prometheus, Cortex, and Loki can be created and managed from a single searchable view. Legacy Grafana alerts will eventually be deprecated.

Grafana 8.0 also adds some bells and whistles including the ability to build “library panels” that can be reused across multiple dashboards. There are also some BI-like features (as Grafana can pull data from relational databases) like State Timeline Charts, which shows the changes in the status of an entity or data point over time; and bar charts — that are being introduced in beta.

These BI-like features have come about by popular demand, like Oracle, PostgreSQL, MySQL, and other relational databases are commonly used data sources for Grafana users. But our concern is that this could become a distraction for the Grafana project, which has set a sweet spot in observability, providing lighter weight, open-source alternative to InfluxDB, Elasticsearch, Splunk and others. Grafana should not attempt to become another Tableau.

On the Grafana Enterprise side, it is adding fine-grained, role-based access controls, and a new data caching feature that could be especially useful when querying popular, and proprietary data sources like Splunk or Oracle. With caching, IOPS costs against the database can be reduced. And the cloud service is also adding support for Grafana Tempo as part of its observability stack of Grafana Cloud. Tempo is an open-source distributed tracing back end that was introduced last fall that can be used with open-source tracing protocols such as Jaeger, Zipkin, and OpenTelemetry.

As noted above, Grafana has carved out a growing niche as an open-source observability engine that has gone viral. We saw that in the readership for our post, a few months back, for AWS’s new Managed Grafana Service that it delivered in a partnership with Grafana Labs. So it makes sense that on the horizon, we may see Grafana penetrate the edge, especially in use cases such as with industrial machinery. In the current release, it is adding support for real-time streaming via a WebSocket connection, where metrics are broadcast to all dashboards connected to the stream endpoint.

With the exception of some of the BI features, Grafana 8.0 and the updates to Grafana Enterprise are generally available now.

Leave a Reply

Your email address will not be published. Required fields are marked *