Published: 14/11/2025
6 minute read
I can’t recommend Grafana anymore
Disclaimer: This tells my personal experiences with Grafana products. It also incudes some facts but your experience may entirely vary and I would love to here your take.
I started my work life at a small software company near my university. They develop, run websites and operate web services for multiple clients. Everyone had multiple responsibilities, and they heavily relied on interns and freshmen—which can be both bad and good.
For me it was good because I learned a lot.
At some point we needed a monitoring solution, and Zabbix didn’t fit well into the new and declarative world of containers and Docker. I was tasked to find a solution. I looked at Loki/Prometheus with Grafana and Elastic with Kibana. Elastic was a beast! Heavy, hard to run, resource-hungry, and complex, Loki and Prometheus were the perfect fit back then.
So I created a docker-compose.yaml with Loki, Prometheus and Grafana. Since they all had the internal Docker network, we required no auth between them. Grafana was only exposed over an SSH tunnel. One static scrape config and the Docker Loki log plugin later, we had our observability stack. For non-Docker logs, we used Promtail. Loki and Prometheus stayed on the same machine and all we required was a local volume mount. Load was minimal.
ℹ️ This is when I learned that you should not transform every log parameter to an label just to make it easier to select in the Grafana UI. Having a label for latency with basically limitless values will fill every disks inodes, thats just how Cortex bin-packs.
I also found out Grafana Labs has a cloud offering with a nice free tier. So I even used this for personal stuff. I had a good experience with them.
... continue reading