Amazon Managed Service for Prometheus: The new Prometheus-based container monitoring service

Get a free Techzine subscription!

Amazon Web Services announced a new cloud service on Wednesday aimed at helping customers monitor the performance of their Kubernetes-managed container software deployments. The service, named Amazon Managed Service for Prometheus, is now generally available.

Customers should find it an easier-to-use solution for monitoring modern applications. The service uses the open-source Prometheus (a well-liked containers monitoring tool) as its basis.

Developers deploy containers to host app components because they are lightweight, agile, and easy to move between different types of computing infrastructure. Prometheus integrates with Kubernetes, the software used to orchestrate many containers.

The problem with hosting apps in containers

Although hosting apps in containers has its advantages, it can get problematic too. For instance, most containers only run briefly and share infrastructure resources with others.

The volume of operational metrics is also large, making it harder to monitor everything using traditional application monitoring which runs on virtual or physical servers. That is why companies often have difficulty getting insights into the health or performance of container apps, which is more challenging, especially when deployed at scale.

If companies lack insight into app performance, they risk getting subpar performance or disruptions.

Prometheus

Many companies turn to Prometheus for help, as it is optimized to collect operation metrics from containers and push alerts when there is a potential problem. This approach introduces new problems when companies scale it across multiple servers, as it is hard to configure for high availability.

On top of that, it takes a lot of time to optimize how Prometheus uses memory and storage resources to limit costs while improving response times.

Amazon Managed Service for Prometheus is the solution debuted to make all this work easier. There is no learning curve (it uses the same data model and query language as Prometheus), making things easier to manage.