The future of Promitor is community-driven

Tom Kerkhove -

In 2018, I was working for a customer that was migrating from Azure Cloud Services to Kubernetes. They used to use Azure Monitor Autoscale to scale their workloads based on Azure Service Bus queues but could not do so with Kubernetes.

At that point in time, the Kubernetes ecosystem did not have KEDA that solves this problem nicely yet so I had to come up with another solution. Prometheus, however, offered an external metrics adapter that the Horizontal Pod Autoscaler could use to autoscale workloads.

Spending enough time traveling back then, I started hacking a POC together to bring Azure Monitor metrics in Prometheus. That POC became an open-source project et voila, and in February 2018 Promitor was created.

In 2019, I was invited by Microsoft to come over to Redmond and do a hackathon because there were so many customers interested in using it they also wanted to get their Azure Monitor metrics in Prometheus (and other sinks in the future).

I've seen (massive) enterprises starting to adopt Promitor who soon had a new problem - Declarative metrics are nice, but they had thousands of resources which was not scaling for them. In 2020, I started designing a resource discovery solution with Maarten Balliauw when I was traveling to Finland.

I am astonished how Promitor has evolved from an idea I had to an open-source product that is adopted by 160+ enterprises, doing 300 million requests per day. (based on the limited telemetry it offers)

Why are things changing?

Well... life happened - I became a proud father of 2 kiddos in the middle of a global pandemic, got a new job and did not need to use Promitor myself anymore.

That resulted in me having less available time to focus on Promitor and keep on evolving the product, although I definately wanted to do so. (for example support the new data-plane API)

What is changing?

Due to the reasons above, I am/have not been able to keep on contributing new features which may have been a point of frustration.

While I have been trying hard to fix this problem, I have to face the fact that I cannot solve it and have to go to a "external contributions" model where I am no longer able to contribute things myself.

Does this mean Promitor is deprecated? That will depend on how many external contributions are going to come in the next year:

  • If there are few, I will officially deprecated and archive the project.
  • If there are more and more contributors, I am happy to work with the community to transfer it to new maintainers inside the Promitor organization

I will, however, still do releases for changes coming in when we have accumulated a decent amount of changes or a security patch was made.

How does this impact you?

In theory, nothing changes. However, it highly depends on how the community is going to contribute.

The best way, as a Promitor user, to ensure that it keeps on evolving is by making contributions to the project.

Summary

I've been thinking long and hard about writing this blog post, but unfortunately there is no other way and I want to be transparent to Promitor end-users about this.

However, it is not dead and the community can still contribute! I just don't have time to actively contribute code anymore, sadly.

Thanks for reading,

Tom