Promscale has been discontinued

[ramonguiu]

Dear community,

We regret to inform you that we’ve made the difficult decision to discontinue Promscale.

Promscale was created to help the community use PostgreSQL/TimescaleDB as a fully compliant remote storage for Prometheus. Later on we added OpenTelemetry and Jaeger tracing support. With its full SQL support, it provided a unified storage for traces and metrics with deep analysis and correlation capabilities. We also built tobs to simplify the deployment of a complete observability stack with one command.

Promscale has been a labor of love for our team at Timescale for over two years, and we’re grateful for the support and feedback from you, our community of users. However, after reevaluating our company priorities, we’ve decided to redirect our resources to other product areas.

We understand this news may be disappointing, so please see the FAQ for answers to common questions including alternative solutions.

Thanks again for your support.

FAQs

Why is Promscale discontinued?
We’re re-focusing all our efforts on the core of what we do best, creating a great developer experience for time-series application developers.

What happens next?
The following Github repositories will remain publicly available but will no longer be evolved nor maintained:
https://github.com/timescale/promscale
https://github.com/timescale/promscale_extension
https://github.com/timescale/tobs
https://github.com/timescale/helm-charts/tree/main/charts/promscale

We’ll continue to provide support for critical issues to existing customers of Promscale with Timescale Cloud or MST until April 30th, 2023, at which point support will end.

Can I keep using Promscale with TimescaleDB?
You can continue to use Promscale with TimescaleDB although we don’t recommend it since Promscale will no longer be supported nor maintained.
The Github repositories and release artifacts will remain publicly available. We can’t ensure that Promscale will continue to work with newer versions of TimescaleDB and PostgreSQL.
If you want to make modifications, we encourage you to fork the repositories.

Can I keep using the open-source version of Promscale with Timescale Cloud / MST?
You can until April 30th. Before April 30th you should transition to a different solution since we can’t ensure that Promscale will continue working and we plan to remove the unsupported promscale extension from our servers.

Who do I contact if I am a customer of Timescale Cloud / MST and I need critical support?
Contact the support team at support@timescale.com or https://www.timescale.com/support

What Promscale alternatives do you recommend?
There are different options depending on your use case.

If you use Promscale as a remote storage for Prometheus there are a number of other options you can consider. There is a list provided in the Prometheus documentation.

If you are using Promscale as a storage system for your OpenTelemetry or Jaeger traces there are a number of options you can consider. This is a list of products that have native support for OpenTelemetry. You can use the OpenTelemetry Collector to convert your Jaeger traces to OpenTelemetry traces.

If you have a specific need, you can ask in the #promscale channel in the Timescale slack community and we’ll try to help you.

[emanzx]

sad.. it was fun using it. sad to see it goes.. but for the best for the team.. all the best..

[felddoteth]

But this is the best storage solution for self-hosted Prometheus metrics 😭

[aantn]

Hey, I'm sorry to see this discontinued, but I understand the need for hard prioritization decisions. It was always a pleasure seeing Promscale at Prometheus Day and KubeCon events.

Just to clarify, is TimeScale shifting away from observability data altogether? Or just away from Prometheus compatibility?

[ramonguiu]

Thanks, @aantn. This announcement only affects Promscale.

Timescale will continue to evolve as a flexible and powerful database that can be used for a number of different use cases.

[Remigius2011]

Hi, we are currently researching to create an observability stack and would like to use TimescaleDB as its storage. However, it seems that now as Promscale is deprecated, there is no part left that would connect either Jaeger / OTEL / Prometheus with Timescale. How could we achieve this?

As a side remark: when searching for connecting OTEL with Timescale, the results are still full of mentions of Promscale. In addition, there are articles in the Timescale blog that refer to Promscale without even mentioning its deprecation. The deprecation notice on the top of the promscale does not mention a sane replacement (neither are the alternatives mentioned above really useful). IMHO deprecating Promscale looks like cutting off Timescale from a multitude of use cases involving OTEL / Jaeger / Prometheus. In our case, we might have to look elsewhere for a storage if we cannot find a replacement for Promscale, even though we have compelling reasons to prefer Timescale to all the other considered alternatives.

[tnn2]

Bummer, we have a significant amount of custom reporting that pulls data out of the hypertables. Not trivial to replace with Mimir. Any chance of a community maintained fork happening?

[dscarabelliTT]

Hey, can anyone suggest a method to connect node exporter to TimescaleDB?