What exactly is Hyper Converged Infrastructure and how does it benefit VDI? The attraction of the computer and technology as a whole has been to make our lives easier. In many ways, business technology has added difficulty and complexity instead. Traditionally when maintaining data center equipment companies would need to purchase a storage system that would be shared by all the workloads, compute resources that would attach to the storage and work on the data, and switching that would connect the compute to the storage resources. This added complexity in that you would have to manage each portion separately. Additionally, in many larger companies there would be silos of resources. Some departments were responsible for only one of the three components, and if you were in one silo you likely couldn’t change anything in the others and usually didn’t have access even if you knew what to do.
The first movement was to create the Converged Infrastructure where the compute, storage, and switching were in a single hardware purchase and in many cases managed from a common console. This made things easier with the common console, but it still had the difficulty of requiring the users to know how to setup each of the components etc.
Then along came the Hyper Converged Infrastructure. In the hyper converged infrastructure, all of the components are software defined and can be deployed on commercial off the shelf equipment. The storage utilizes locally attached storage on the servers and creates a virtual SAN. In many cases, the switching between the servers can also be software defined permitting the single console control and ease of management desired.
In Hyper Converged Infrastructure, resources are purchased in building blocks. You have a known quantity of compute, memory, and storage in each node, and when more is needed, you can add a set amount. Many instances now do permit the addition of compute only and storage only if there is a specific shortage of a particular resource. But, the software defined piece is maintained.
One of the attractions to running a Virtual Desktop Infrastructure (VDI) on Hyper Converged Infrastructure (HCI) is that VDI has a known quantity of resources required for support. Most HCI vendors can give you a reference architecture that shows how many users per node could be supported for a VDI workload. When you reach the user threshold, you add another node and so on. This can be very attractive to companies that utilize VDI.
I recently interviewed a couple of customers that have deployed different HCI vendor’s products to support their VDI environments. We had a virtual meeting (Thanks COVID) and discussed how they were using the environments, and how they felt the design was performing.
First, I spoke with Jamie Terrell, a Citrix Engineer supporting the Florida Cancer Specialists VDI deployment. He supports several thousand employees across the country utilizing Citrix Apps and Desktops on a regular basis. They have chosen Nutanix as their vendor for the environment. Currently, they are using the Acropolis Hypervisor and only VDI runs on the Nutanix HCI. They are running well and Jamie said he is warming up to the Acropolis Hypervisor (AHV) more as he gets to use it. They definitely benefit from the ease of adding in resources and controlling their workloads that HCI offers. They are expanding their environment to include App Layering and they have used Nutanix Files to ease some of the burden on VDI as well
Next, I spoke with Shan Venable, the CTO and SVP at Altamaha Bank and Trust. Shan had experience with Hyper Converged Infrastructure and Citrix VDI at his former employer and knew he wanted to go in that direction at Altamaha. He looked at Nutanix but decided that SimpliVity fit better. He liked the dedupe on the fly and the ability to be redundant with two nodes instead of three. They run their entire infrastructure on HCI. When it was initially brought in, it was before they had been purchased by HPE. They spun up test servers and were able to see that ran correctly, Shan even caused a physical failure of one node (unplugging) and everything failed over and ran as expected. In about 4 months they had migrated all their workloads to the new infrastructure. They have nodes in DR as well, which they have failed over, ran for a couple days and failed back, no issues. Last year they upgraded to the new DL380 All Flash Gen 10 versions and the workloads are much faster now. With SimpliVity dedupe and compression they are getting about 22:1 on their storage. It runs the VDI environment without issue.
With the ease of use, and ease of configuration with VDI providers, Hyper Converged Infrastructure seems to be a perfect fit for VDI. You can find out more information on the platforms here in the Gartner Magic Quandrant for HCI. Of course, you can reach out to me and I would be glad to assist in helping you to determine if HCI is right for you.
Case Study Altamaha Bank and Trust
Reference architectures for Citrix VDI
Reference Architectures for VMware Horizon View