Why Grafana Labs delivers the best Prometheus in the Cloud

Over the last several months, there have been a variety of service providers that have launched Managed Prometheus offerings. This is a testament to the rise in popularity of the Prometheus project, and how it’s becoming a de facto standard for metrics.

The most recent announcement in the Managed Prometheus landscape came from AWS. During a re:Invent keynote on Dec. 15, Amazon announced its AWS Managed Service for Prometheus. The service is built using the CNCF’s Cortex project, the open source, horizontally scalable Prometheus-compatible project that I started with Julius Volz over four years ago. I’d like to take this opportunity to extend them a warm welcome to the Prometheus-as-a-Service club! We think AWS made a good choice choosing Cortex, and see this as a massive vote of confidence in the project.

Grafana Labs, on behalf of AWS, made improvements to the open source Cortex project such as extending the scalability and isolation it offered. As part of our engagement with AWS, we added “shuffle sharding” (a way of improving tenant isolation and scalability) into Cortex, built a new Query Scheduler (to improve the scalability of the Query Frontend component), and enabled Cortex to tolerate a complete Availability Zone failure. Working closely with the AWS team has been an absolute pleasure.

We are heavily invested in the success of the Cortex project and any user that adopts it. The scalability testing, input, and review the AWS team performed have been invaluable to the rapid rollout of these new features.

Now that you can get Hosted Prometheus from AWS, why would you use Grafana Cloud? We think the key differences break down into three categories:

Grafana Cloud is more than Cortex: We add numerous features on top of the open source Cortex project, such as:

• fine-grained access control, which allows you to restrict who can access which metrics;
• a sophisticated integrations wizard that packages up Dashboards, Alerts, and Exporters in a way that makes it quick and easy to monitor your target system;
• deep integration with other Prometheus-inspired approaches to log aggregation with Loki and tracing with Tempo;
• a dedicated Grafana instance, with a native integrated Prometheus Alerts and Rules UI.

Grafana Cloud can be simpler and more cost-effective: Our pricing structure only considers write volume, and can cost over 5x less than AWS for a comparable workload. We also offer longer retention (13 months), and with the Grafana Agent, a single thing to deploy and manage — simplifying your Prometheus journey. We’ve just announced new free and paid Grafana Cloud plans to suit every use case — sign up for free now.

Grafana Labs are the Prometheus and Cortex experts: Grafana Labs’ employees started the Cortex project and have been driving it for over four years; we employ five of the eight Cortex maintainers and have over a decade of collective experience running Cortex. We also employ 6 of the 17 Prometheus maintainers, ensuring these two projects stay closely aligned. We’ve helped numerous organizations adopt Prometheus, learned the hard lessons, and seen the common pitfalls we can help you avoid.

The recent explosive growth of companies offering hosted Prometheus only makes it easier for you to choose a diverse set of service providers for your observability stack, and avoid correlated failures. We’re excited to enter this new phase for the Cortex and Prometheus projects and look forward to collaborating with AWS for years to come.  

P.S. Want the best-of-breed solution for scaling Prometheus, but can’t use the cloud? We also offer Grafana Enterprise Metrics, bringing all the features of Grafana Cloud to on-premise or self-managed environments.