# New ES cluster: node type configuration

**URL:** <https://discuss.elastic.co/t/new-es-cluster-node-type-configuration/63975>\
**Category:** Elasticsearch\
**Created:** [October 26, 2016, 9:01am UTC](https://discuss.elastic.co/t/new-es-cluster-node-type-configuration/63975 "2016-10-26T09:01:56Z")\
**Posts on this page:** 4\
**Page:** 1

<div class="post-metadata">

**Author:** ![andreas\_at\_work](https://avatars.discourse-cdn.com/v4/letter/a/e9c0ed/32.png) [@andreas\_at\_work](https://discuss.elastic.co/u/andreas_at_work)\
**Post date:** [October 26, 2016, 9:01am UTC](https://discuss.elastic.co/t/new-es-cluster-node-type-configuration/63975/1 "2016-10-26T09:01:56Z")

</div>

Hey,

after the [descision](https://discuss.elastic.co/t/new-cluster-hardware-descision/61918) was made and the hardware was ordered we have to make the next step:

We ordered the following hardware:  
8x ES node

- CPU 2620v4
- Cores 8
- Cache 20MB
- Freq 2,1GHz
- RAM 64GB
- HD 2x960GB SSD

What kind of node configuration would you recommend?

1. 8x equal nodes (all master and all data)?
2. 3x master and 5x data nodes
3. 1x master and 7x data nodes

How much of index/search performance we do might loose with each config?

Client nodes are out of scope and will be operated on other hardware.

Thanks  
Andreas

---

<div class="post-metadata">

**Author:** ![warkolm](https://sea2.discourse-cdn.com/elastic/user_avatar/discuss.elastic.co/warkolm/32/39224_2.png) [@warkolm](https://discuss.elastic.co/u/warkolm)\
**Post date:** [October 26, 2016, 9:12am UTC](https://discuss.elastic.co/t/new-es-cluster-node-type-configuration/63975/2 "2016-10-26T09:12:45Z")

</div>

> [@andreas\_at\_work](#):
>
> How much of index/search performance we do might loose with each config?

You'd need to test that yourself.

---

<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:** [October 26, 2016, 9:12am UTC](https://discuss.elastic.co/t/new-es-cluster-node-type-configuration/63975/3 "2016-10-26T09:12:55Z")

</div>

Dedicated master nodes do very little work, so having a full host for just a master node would be wasting resources. Each server should therefore have a node that holds data. As you always want at least 3 master eligible nodes in a cluster, having all of them be master eligible is fine as long as you set minimum\_master\_nodes set to 5. You can also choose to have only 3 of them be master eligible or possibly even host a separate small dedicated master node on 3 of the servers.

---

<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 5, 2017, 10:09pm UTC](https://discuss.elastic.co/t/new-es-cluster-node-type-configuration/63975/4 "2017-07-05T22:09:40Z")

</div>