Prometheus is an open source monitoring tool that was originally built by SoundCloud in 2012, inspired by Google's BrogMon. It is written in GoLang. According to the New Stack Survey of 2017, Prometheus is one of the most widely used tools for monitoring Kubernetes clusters. What makes Prometheus different than other open source monitoring systems is that it has a simple, text-based format, making it easy to get metrics from other systems. It also has a multidimensional data model and a rich and concise query language. Using Prometheus, we can monitor all levels, nodes, container-scheduling systems, and also routers and switches. If we are dealing with large applications and a fast-moving infrastructure, this means that the jobs that we run change rapidly and we have to deploy them around 100 times a day. In this case, Prometheus will be very useful, as it has the ability to discover services. If we have a dynamic infrastructure, we can use Prometheus to detect early failures and determine what's going on in the entire stack. It also helps developers to investigate how and why something went wrong. Prometheus is not, however, very good for producing logs.
Let's talk about the positive features of Prometheus:
- It has a multidimensional data model that uses time-series data. It uses a pull model to fetch time-series data over HTTP.
- It has a powerful and flexible query language. This can be useful when setting up graphs and monitoring dashboards.
- It is easy to integrate with other tools, such as Grafana.
- It is easy to investigate the application failures.
Prometheus, however, does not include the following:
- Raw log/event collection
- Request tracing
- Anomaly detection
- Durable long term storage
- Automatic horizontal scaling
- User authentication management