I have explored couple of observations and wanted to share seek necessary 
guidance 

One cons i see with Shards is -  if you have reached a limit on max targets 
(i.e total scrape time taking longer than intended scrape interval) or if 
you can’t vertically scale the pod/machine anymore.
Sharding complicates your setup (you need more component to get a unified 
view), adds to query latency (due to fanout) and increases your costs.

In that case  - Do we have any approach/methodology of determining the 
“number of shards” based on the and number of targets/scrapping ?



On Sunday, August 30, 2020 at 7:46:33 PM UTC+5:30 [email protected] 
wrote:

> Can someone just share the knowledge about trades on sharding Prometheus 
> into multiple nodes
>
> Any hints/pointers would definitely be helpful
>
> On Fri, 28 Aug, 2020, 12:31 pm [email protected], <
> [email protected]> wrote:
>
>> Hi Team,
>>
>> we have implemented huge number of shards in prometheus which scrapes 
>> 1000 target nodes across the enterprise 
>>
>> we are currently in process of running some load test and ascertain the 
>> unknowns one below and need some guidance on the same - 
>>
>>
>>    1. On what cases sharding approach can fail?
>>    2. How do we quantify the load/scale requirements?
>>    3. Is there any optimized settings/configurations or best practices 
>>    we need to comply for sharing 
>>    4. Any low level tuneable like recording 
>>    rules/metrics_relabel_configs/scrape_internal and sample we ingest we 
>> need 
>>    to accomodate and any pointers to it ?
>>
>> Please do guide us through whether we think on right page and correct me 
>> incase of any misinterpretation
>>
>> Regards,
>> Dinesh
>>
>> -- 
>> 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 [email protected].
>> To view this discussion on the web visit 
>> https://groups.google.com/d/msgid/prometheus-users/3307c7af-6fac-4e5a-a422-d2eee13c9fccn%40googlegroups.com
>>  
>> <https://groups.google.com/d/msgid/prometheus-users/3307c7af-6fac-4e5a-a422-d2eee13c9fccn%40googlegroups.com?utm_medium=email&utm_source=footer>
>> .
>>
>

-- 
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 [email protected].
To view this discussion on the web visit 
https://groups.google.com/d/msgid/prometheus-users/000f64ca-149b-48f1-8846-416182ae3f5en%40googlegroups.com.

Reply via email to