# Indexing speed and SSD hardware

**URL:** <https://discuss.elastic.co/t/indexing-speed-and-ssd-hardware/216720>\
**Category:** Elasticsearch\
**Created:** [January 27, 2020, 8:03pm UTC](https://discuss.elastic.co/t/indexing-speed-and-ssd-hardware/216720 "2020-01-27T20:03:47Z")\
**Posts on this page:** 6\
**Page:** 1

<div class="post-metadata">

**Author:** ![tlandelin](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/tlandelin/32/61497_2.png) [@tlandelin](https://discuss.elastic.co/u/tlandelin)\
**Post date:** [January 27, 2020, 8:03pm UTC](https://discuss.elastic.co/t/indexing-speed-and-ssd-hardware/216720/1 "2020-01-27T20:03:47Z")

</div>

In [Tune for indexing speed | Elasticsearch Guide [master] | Elastic](https://www.elastic.co/guide/en/elasticsearch/reference/master/tune-for-indexing-speed.html) it says

> If indexing is I/O bound, you should investigate giving more memory to the filesystem cache (see above) or buying faster drives. In particular SSD drives are known to perform better than spinning disks....

I've been experimenting with indexing with SSD drives, but I'm not finding significant improvement in indexing speed. I'm left to conclude that likely my indexing is not I/O bound.

How can I tell for sure whether indexing is I/O bound?

Also, are there any recommended settings for indexing when backed by SSD drives?

---

<div class="post-metadata">

**Author:** ![Christian\_Dahlqvist](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/christian_dahlqvist/32/4617_2.png) [@Christian\_Dahlqvist](https://discuss.elastic.co/u/Christian_Dahlqvist)\
**Post date:** [January 27, 2020, 8:21pm UTC](https://discuss.elastic.co/t/indexing-speed-and-ssd-hardware/216720/2 "2020-01-27T20:21:21Z")

</div>

Are you using locally attached SSDs? What does disk utilisation and cpu usage look like during indexing?

---

<div class="post-metadata">

**Author:** ![tlandelin](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/tlandelin/32/61497_2.png) [@tlandelin](https://discuss.elastic.co/u/tlandelin)\
**Post date:** [January 27, 2020, 11:08pm UTC](https://discuss.elastic.co/t/indexing-speed-and-ssd-hardware/216720/3 "2020-01-27T23:08:58Z")

</div>

Yes, SSDs are locally attached.

With our regular HDD cluster, during indexing (using sar to measure), CPU averages 15% and is blocked by disk I/O on average 1.5% of the time. Measurements include search traffic.

---

<div class="post-metadata">

**Author:** ![rugenl](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/rugenl/32/12887_2.png) [@rugenl](https://discuss.elastic.co/u/rugenl)\
**Post date:** [January 27, 2020, 11:46pm UTC](https://discuss.elastic.co/t/indexing-speed-and-ssd-hardware/216720/4 "2020-01-27T23:46:17Z")

</div>

Indexing should be "write" bound before "read" bound, memory cache may not help write rate.

Are these SATA or SAS SSD? What OS and filesystem?

---

<div class="post-metadata">

**Author:** ![tlandelin](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/tlandelin/32/61497_2.png) [@tlandelin](https://discuss.elastic.co/u/tlandelin)\
**Post date:** [January 28, 2020, 12:37am UTC](https://discuss.elastic.co/t/indexing-speed-and-ssd-hardware/216720/5 "2020-01-28T00:37:32Z")

</div>

Disks are SAS. Running debian linux and using xfs

---

<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:** [February 25, 2020, 12:46am UTC](https://discuss.elastic.co/t/indexing-speed-and-ssd-hardware/216720/6 "2020-02-25T00:46:05Z")

</div>

This topic was automatically closed 28 days after the last reply. New replies are no longer allowed.
