Objects Scale-out : breaking physical boundaries.

Anirudha | Wed, 08/19/2020 - 08:25

Nutanix stack is built on distributed architecture and is highly scalable & availability. You can scale Nutanix AOS cluster, by adding more disks in the node Or  nodes in the cluster Or blocks across the rack Or even stretch clusters across the multiple racks. Each of these actions gives you the next level of high availability and scalability.  Flexibility that Nutanix AOS stack provides is just beyond amazing.

But when we talk about Nutanix Objects, it does something even more amazing. In this blog we will go over how scalable Objects cluster is.

What you need :

  1. 4 Node AOS cluster and access to its prism central server.
  2. Access to PrismCentral for deploying Objects and further experiments.

Quick look at AOS Cluster and Prism Central :

This is a typical 3N AHV cluster, registered with Prism Central. This cluster has around 23.18TB physical space and ~21.24TiB is free.

I have deployed 3N Objects cluster on this.

This PC has 4 Objects clusters deployed on different AHV clusters. And we are going to work on InstanceName : Test-Drive for today’s experiments.

As you can see I have about 23TiB storage in my AOS cluster, and quick look at what Objects reports :

If you navigate to the Instance UI and usage page, it shows the same stats , 23.18TiB storage usable and it has already consumed, 1.85TiB.

In the first screenshot you can see Test-Drive instance level stats reports ~938GiB usage while above screenshot reports ~1.85TiB. That is because , ~938GiB is logical space used by Objects cluster i.e user has uploaded ~938GiB of data and but on the underlying cluster, it has consumed ~.185TiB physical space (refer AOS documentation to understand Logical vs Physical storage usage.)

By default Objects will use all available resources from the underlying AOS cluster. But in case you want to restrict the storage usage then refer to the “Limiting or Updating Storage” post to achieve the same.