The hardware required of Promethues

[Mạnh Nguyễn Tiến]

Dear All,

I am calculating the hardware requirement of Prometheus. Just minimum hardware requirements. I found some information in this website: https://prometheus.atlas-sys.com/display/Ares44/Server+Hardware+and+Software+Requirements. But I don't know that it is for how many node and how many metric and sample. Please help me. Thanks.

[Ben Kochie]

I don't think that link has anything to do with Prometheus. :)

For the most part, you need to plan for about 8kb of memory per metric you want to monitor.  A typical node_exporter will expose about 500 metrics.

100 * 500 * 8kb = 390MiB of memory.

There's some minimum memory use around 100-150MB last I looked.

On Mon, Sep 17, 2018 at 9:32 AM Mạnh Nguyễn Tiến <tienma...@gmail.com> wrote:

Dear All,

I am calculating the hardware requirement of Prometheus. Just minimum hardware requirements. I found some information in this website: https://prometheus.atlas-sys.com/display/Ares44/Server+Hardware+and+Software+Requirements. But I don't know that it is for how many node and how many metric and sample. Please help me. Thanks.

--
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 post to this group, send email to promethe...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/prometheus-users/54d25b60-a64d-4f89-afae-f093ca5f7360%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Ben Kochie]

Sorry, I should have been more clear. I menat to say 390 + 150, so a total of 540MB.

These are just estimates, as it depends a lot on the query load, recording rules, scrape interval.

Here's a typical example: sum(process_resident_memory_bytes{job="prometheus"}) / sum(scrape_samples_post_metric_relabeling)

On Tue, Sep 18, 2018 at 5:11 AM Mạnh Nguyễn Tiến <tienma...@gmail.com> wrote:

Why the ressult is 390MB, but 150MB memory minimun are requied by system.

Best regards,

Vào Th 3, 18 thg 9 2018 lúc 04:32 Ben Kochie <sup...@gmail.com> đã viết:

Yes, 100 is the number of nodes, sorry I thought I had mentioned that.

Prometheus is a polling system, the node_exporter, and everything else, passively listen on http for Prometheus to come and collect data.

This allows for easy high availability and functional sharding. The exporters don't need to be re-configured for changes in monitoring systems.

On Mon, Sep 17, 2018 at 7:09 PM Mạnh Nguyễn Tiến <tienma...@gmail.com> wrote:

Thank you so much. Can you describle the value "100" (100*500*8kb). Is it number of node?. Btw, node_exporter is the node which will send metric to Promethues server node?

Vào Th 2, 17 thg 9 2018 lúc 22:53 Ben Kochie <sup...@gmail.com> đã viết:

--

Nguyen Tien Manh

Hanoi University of Science and Technology

Tell: +84 166 557 6448

--

Nguyen Tien Manh

Hanoi University of Science and Technology

Tell: +84 166 557 6448