Prometheus - Thanos vs VictoriaMetrics for HA and storage retention

[Bruno Cochofel]

Hi,

I was looking at installing Thanos along with Prometheus for HA and storage retention and found VictoriaMetrics.

Seens easier to install/configure. Does anybody has experience with any one of these tools that can help on witch to choose?

From what I've read, VictoriaMetrics seens a better fit in just about everything.

Thanks,

Bruno

[Aliaksandr Valialkin]

I think it would be better asking this question on our Slack channel - http://slack.victoriametrics.com .

--
You received this message because you are subscribed to the Google Groups "Prometheus Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to prometheus-use...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/68d3ad63-3601-4d31-98c6-b03922e9dbfe%40googlegroups.com.

--

Best Regards,

Aliaksandr

[Chris Bulleri]

I'd love to hear about any lessons learned that you may have.  I'm not ready to implement either solution but would love to learn from you.

[Aliaksandr Valialkin]

The major difference is that Thanos stores data on object storage such as S3 or GCS, while VictoriaMetrics stores data on persistent storage (either local or network-attached such as EBS or GCP disk). Both approaches have pros & cons, which are described at  https://medium.com/faun/comparing-thanos-to-victoriametrics-cluster-b193bea1683 .

--

You received this message because you are subscribed to the Google Groups "Prometheus Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to prometheus-use...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/a0940aeb-9548-4046-a003-5ddd0a32207d%40googlegroups.com.

--

Best Regards,

Aliaksandr

[Bartłomiej Płotka]

Just to clarify it's worth to note that there other significant differences:

* You can configure Thanos to NOT use Prometheus remote-write completely and I think that's the simplest option currently. Instead, Thanos is capable of streaming the metric data from Prometheus directly which allows smooth, gradual installation process - you just add sidecar plus single service. Then you can move to more complex setup with object storage for long term retention or even remote write.  
* Thanos uses native Prometheus/TSDB format in the object storage. So native analysis, compaction, debugging Prometheus tools work just fine.

Note that Thanos on top of GCS, S3 also supports Azure, Tencent, OpenStack Swift and more. 

The last thing to remember is that both I and Aliaksandr might be biased (: I am Thanos co-author, Aleksandr founded VictoriaMetrics. 

As always I would recommend to try both (including another option: Cortex) and gather more input on what fits your needs the most.

Kind Regards,
Bartek

To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/CAPbKnmB608uHNBj3UQMZx1Ya4%2B2oFYXNmnw4WVRv-yUEZkgFZA%40mail.gmail.com.