S3 object store over multiple clusters with single namespace | Nutanix Community
Skip to main content

L.s,

First time here with a question  (there must be a first time)

We want to create a environment with a Splunk/Nutanix combination. The storage for Splunk will be smartstore, so the warm storage is S3 compliant storage, something Nutanix can provide.

The question comes from the fact that we will span Splunk acros two site’s. So each site will have a storage cluster and a compute cluster (multiple clusters because more then 3 hops).  Site A has a compute cluster and a storage cluster, site B has the same. The Storage cluster will be synced between eachother.

Splunk validated architecture says to have a loadbalancer between the Splunk indexers and the S3 storage. So when the storage in Site B fails.. Site B can use Site A as storage. But also the loadbalancer sends the compute cluster to the nearest storage available.

Does Nutanix has something for this loadbalancer question? Can two clusters be adressed  by one ip, ore is it indeed necessary to use a external loadbalancer?

Thanks in advance for the answer :)

Greetings

Jari

 

 

 

Hi Jari,

 

With Nutanix Objects v 4.0 a new concept has been introduced as federation namespace, “In the context of a single Objects instance, namespace is a set of unique buckets hosted by that instance. Federation provides a way to create a single namespace across multiple individual Objects instances. One federation instance corresponds to one such namespace.”

https://portal.nutanix.com/page/documents/details?targetId=Objects-v4_0:top-federation-overview-c.html

The above link might give better idea, and you are always welcome to ask your Nutanix AM/SSE.

 

F>P

Thanks :)

 

This seems the right way. I will take this to the designers for a test .

 

greets

Jari

Terms & Conditions