Storage is perhaps the most important ingredient in virtual desktop infrastructure deployments, but calculating how much a virtual infrastructure needs isn't easy. Hyper-converged infrastructure changes that.
In a virtual desktop infrastructure (VDI) deployment, the storage infrastructure must provide sufficient space to accommodate the virtual desktops, but VDI storage performance is almost always the bigger issue. Administrators must ensure that storage hardware will keep pace with VDI's demands.
Many organizations find that storage performance issues in VDI deployments creep up over time as the company grows. IT typically invests time and effort in the initial VDI planning process to ensure a good end-user experience, but as demand on the virtual infrastructure increases, the underlying storage might not support the increased demand for IOPS.
There are a variety of ways to squeeze better performance out of the storage infrastructure, but they aren't always easy or cost-effective. First, the organization must determine where the performance bottleneck originated. The Fibre Channel connection could be saturated, or the disks could be pushed to their IOPS limit. In any case, IT admins must find a solution to the problem of their VDI storage performance.
In these situations, the disks are often incapable of collectively delivering the required number of IOPS. A common solution is to install additional hard disks into the storage array, even if additional storage capacity is unnecessary. The added disks are solely there to increase the number of IOPS that the system can handle, and the extra storage capacity is essentially wasted.
HCI to the rescue
Vendors bill hyper-converged infrastructure (HCI) as the platform of choice for VDI deployments, mainly because the appliance is tuned to support VDI's storage needs.
HCI can eliminate the storage scalability problem of VDI deployments because a basic HCI architecture enables modular scalability. Generally, HCI provides modular compute, network and storage resources.
The compute resources within an HCI deployment typically act as cluster nodes. Each node has its own built-in storage resources, enabling the hyper-converged architecture to address VDI storage performance considerations.
From a hardware standpoint, the nodes within an HCI deployment are identical to one another. This makes it easy for an administrator to do VDI capacity planning. If, for example, an administrator were to determine that a node could comfortably run 20 virtual desktops, then a four-node system would be capable of hosting 80 virtual desktops. And because HCI is based on standardized, self-contained modules, the architecture enables linear scalability.
A normal failover cluster does not support linear scalability. Although administrators might be able to accommodate additional virtual desktops by adding nodes to a failover cluster, the cluster's reliance on shared storage means that storage IOPS will eventually become a performance bottleneck.
But, in the case of HCI, each node has its own storage resources. As such, administrators never have to worry about storage performance diminishing, so long as they limit the total number of virtual desktops that are allowed to run on each node. When the organization must accommodate additional virtual desktops, IT can simply add more nodes to the HCI cluster. This reduces the complexity that is so often associated with scaling VDI.
Although HCI does not completely eliminate the need for an administrator who knows something about storage and virtualization prior to deploying VDI, it does reduce deployment complexity. More importantly, HCI enables easy VDI deployment scaling, without negatively affecting VDI storage performance.
Ways HCI can help solve storage problems
Top tips on choosing and implementing HCI
Scalability is one benefit of hyper-converged infrastructure