Prometheus

The Best thing I liked about Prometheus is that It comes with it's own Prom Query Language. Once we learn Prom Query Language, we can write any query to define alerting rules. Also it comes with many aggregate functions which can be used to filter and apply mathematical calculations on the alerts. It comes with wide variety of alerting metrics. Example for Virtual machines and also for Kubernetes Clusters.

There is very minimum things to said as bad, but one thing that i found is you will have to know all the metric names in order to create rules and there are a lot of metrics. Also another thing which i disliked is it doesn't have a stable client library for most of the programming languages.

If you want a monitoring system which is open source then Prometheus is the best choice for you. As it is open source, there is plenty of room for new features. It can incorporate with many alerting systems, which is a plus.

I am using Prometheus to monitor Kubernetes clusters and virtual machines and create alerts based of different situations. The most important benefit i have realized is it supports both webhook and email alerts.

Its reliable and free to use. There is great community support. Fully configurable to pull data from various sources via API calls and rate at which the data is added can be controlled.

Adding system metrics data every minute from across the cluster and dashboarding using Grafana.

Try it ! Best monitoring solution in 2020.

Auto discovery for auto scaled machine in AWS for example.

Nagios was our first monitoring product. We went with Nagios XI for the same reason people buy Cisco switches - nobody ever got fired for buying Cisco hardware, and nobody ever got fired for going with Nagios for monitoring. However, as we grew, we found that Nagios XI doesn't really scale. In addition, both the checks and agent were community maintained and often poorly maintained at that.

We went from Nagios to Sensu - specifically, Sensu Go, which is a greatly improved rewrite of the original codebase. Sensu Go has a really interesting architecture - it's got its own built-in mechanism, Assets, for distributing checks, and the agents are maintained by the company. But the documentation and web UI were terrible, and you were still dependent on writing all checks yourself or sourcing old Nagios checks from the community.

After Sensu more than doubled its support costs year-over-year, we decided that we'd had enough, and I finally gave Prometheus a try. My DevOps friends had been raving about how well it's designed, now I can see why. It's trivial to set up; there's a vast array of strongly-supported exporters, and it's far more flexible than simply checking whether something is running or not.

And it does scale, either through sharding instances or through other projects like Thanos. Nagios is waaay over the hill, and it shows. Sensu isn't much further behind unless they can get their act in order. I see positive things for us and Prometheus.

Use it as no-sql database for run-time data

Using for database for storing run-time data.