# Cpu load on small index small request per second

**URL:** <https://discuss.elastic.co/t/cpu-load-on-small-index-small-request-per-second/8810>\
**Category:** Elasticsearch\
**Created:** [August 22, 2012, 9:40am UTC](https://discuss.elastic.co/t/cpu-load-on-small-index-small-request-per-second/8810 "2012-08-22T09:40:56Z")\
**Posts on this page:** 2\
**Page:** 1

<div class="post-metadata">

**Author:** ![Bastien\_Quelen](https://avatars.discourse-cdn.com/v4/letter/b/58f4c7/32.png) [@Bastien\_Quelen](https://discuss.elastic.co/u/Bastien_Quelen)\
**Post date:** [August 22, 2012, 9:40am UTC](https://discuss.elastic.co/t/cpu-load-on-small-index-small-request-per-second/8810/1 "2012-08-22T09:40:56Z")

</div>

Hi

We use ES for our website with ~260 000 documents on 5 shards, which  
represent approximately 915mo, on a 0.19.8 ES install.  
We made request for fuzzy search with facet, and some facet filters with  
filtered facets inside, the result was fast.  
We get between ~30 request per seconds.  
We begin with freebsd but get some high load like 7 on a 12 core server.  
We try to change JVM to openjdk6-b25\_1 and upgrade lucene to 3.6.1, but  
with no success.  
We add a debian server, we get between 1 and 1.5 load on the debian and 3  
to 5 on the freebsd.  
Both server have JVM memory configurate to 3go and never saturate, and we  
have no cache eviction.  
I add some bigdesk screenshot.

Is there minimal configuration parameters for production server we can have  
miss ?

Thanks  
Bastien Quelen

--

---

<div class="post-metadata">

**Author:** ![system](https://us1.discourse-cdn.com/elastic/original/3X/1/a/1ac57faf039f6b580b3f104ef42a2a89e41014de.png) [@system](https://discuss.elastic.co/u/system)\
**Post date:** [July 6, 2017, 3:15am UTC](https://discuss.elastic.co/t/cpu-load-on-small-index-small-request-per-second/8810/2 "2017-07-06T03:15:31Z")

</div>

Closed on July 6, 2017, 3:15am UTC