Grafana monitoring

• Tanyo Ivanov
• Dec 18, 2022

Hello,
In this post I will show you how I analyze and view my server statistics. It was a long selection process and I went through quite a few analysis tools before choosing this one. At first I knew that Grafana is a good tool and thought about it, but I didn't want to become trivial and decided to try Netdata, DataDog, then I tried Checkmk, then WebMin, then others, but in the end I decided that Grafana is the tool, which gives me the most freedom to change the view and the largest set of statistics, although I think Netdata gives quite a lot out-of-the-box instead of manually creating them yourself. Here you can see a picture of my main Grafana workspace. This is my home page and shows resource usage. I have others that show the use of containers etc:

This workspace has statistics on CPU usage, memory, hard disk, CPU temperature, other resources, SWAP memory, and server downtime. There are some things I don't like frankly about Grafana. The first is that I don't get many things ready-made and I have to do everything myself or download a dashboard from the internet and modify it for my needs. The second thing, and the most important for me, is that parsing itself uses a lot of resources and is a bit of an overhead. In addition, when a dashboard is created, we have to use different datasets, which also allocate certain memory from that of the machine. Node_Exporter uses a lot of memory and I will look into reducing it in the future. I use another desktop that gives me statistics about Docker and the containers in it. From there I learned about memory allocation by the various exporters. It gives me a clearer view of container performance, and based on that I can decide who to allocate how much memory to, and if it's a hand-built system, where to improve it. Here is the second work environment:

Overall, Grafana has its pros and cons. I have used quite a few other tools, such as Microsoft's AppInsights, which gave me pretty good stats and didn't hurt productivity. But that's if we're using Azure, and I'm currently on a Docker host and self-hosted. I also can't see yet a place to get Alerts notifications and detailed system error data like DataDog has. I used DataDog at first, but I don't want the systems data to be kept in foreign companies, so I feel like I don't own the data, it's being used by other people, especially if they're analyzing data. Overall, Grafana gives quite a bit of freedom, privacy and beautiful presentation of the data. I hope you enjoyed what I wrote, I tried to save as much detail as possible. Thank you for reading, See you soon, Tanyo