«

»

A new Nutanix Appliance – All Flash Appliance

As many of you already know, Nutanix web-scale converged infrastructure is a scalable virtualization platform for cloud, desktop, and business-critical applications and big data deployments. I really believe the sentence can be change to:

Nutanix web-scale converged infrastructure is a scalable virtualization platform for all workloads

The reason is our latest announcement, 2014-10-09, of the NX-9000 series appliance which is an all flash appliance. This means Nutanix can now handle extremely demanding workloads with large working sets.

The NX-9040 appliance includes:

  • 2 x Intel Ivy Bridge E5-2690v2 [20 cores / 3.0 GHz]
  • 256 or 512 GB RAM
  • 6 x SSD (800 GB or 1.6 TB)

This means you got approximately 9.6 TB RAW flash storage locally attached to the hypervisor running your virtual machines. The NX-9000 appliance can fit 2 Nutanix nodes meaning you potentially got 19.2 TB RAW flash storage available in 2U.

Screen Shot 2014-10-13 at 11.21.39

The Nutanix scale out solution with one storage controller, controller virtual machine (CVM), per Nutanix node has many  advantages compared to dual flash array available today. Since running one CVM per Nutanix node (hypervisor host) we keep on adding the most critical component for a flash array, processing power, each time we add more capacity (Nutanix node/hypervisor host) to our Nutanix Cluster.

One additional thing, when designing your all flash array system you most likely have to include availability and performance calculations for potential maintenance and/or potential failure.
In that case, when using a dual controller all flash array system, you cannot use more that 50% of the performance of your all flash array during normal operation to be certain that you won’t lose performance in case of one controller becomes unavailable.
If we got 8, 16 or 32 Nutanix nodes in our Nutanix cluster and design for a N+1 failover scenario we can use 87,5%, 93,75%, 96,86% of the available processing power on a daily basis without suffering from performance loss when a Nutanix node becomes unavailable. In the case our Nutanix cluster includes 32 Nutanix Nodes we might use a N+2 calculation giving us 93,75%.

Learn more about all the Nutanix appliance available on the Nutanix web page here.

Below are some additional posts regarding the Nutanix All Flash Array i recommend you to read: