Deploying VDI is a major change in the way desktops are delivered. A “big bang” rollout is usually a recipe for disaster; a cautious approach has a much greater chance of success. A great way to start a VDI project is with a minimum viable VDI platform, which is the smallest VDI deployment you can build in order to see how VDI will work in the wild with your actual end-users. One of the reasons is that users are seldom doing what IT thinks they are doing. Users do what they need in order to get their job done, and it may not be what IT or management expects. Starting slowly means that any miscalculation can be adjusted before it impacts too many staff.
The challenge is to build this minimum VDI in a way that represents how you will build the full VDI deployment. This is where a hyperconverged solution suits VDI so well. Storage performance scales exactly the same way as CPU and RAM—the same amount of performance is added with each hyperconverged node. A hyperconverged node that supports 200 concurrent users can be combined into a cluster for 600 users, or multiple clusters for 6,000 users. More users means more nodes which in turn means more resources. There is no possibility that the storage will run out of performance when you cross some user count threshold.
With storage array-based VDI, running out of storage performance during rollout is a nightmare scenario, but a real possibility. Hyperconverged offers the same performance for each user from the proof of concept (PoC) all the way through the production rollout.
Hyperconverged also suits a careful rollout. Additional nodes can be purchased as users are added. There is no need to buy ten thousand users’ worth of VDI infrastructure at the start of a two-year rollout—nodes can be bought every month or two as the rollout proceeds. Buying in stages minimizes the risk of buying too much or too little hardware, and the ability to add nodes to an existing hyperconverged cluster without downtime allows growth in small units. Naturally, whole hyperconverged clusters can be bought and provisioned at once for faster growing VDI deployments.
Most VDI deployments do not deliver homogenous desktops, because some users require a simple desktop and others require a larger and more complex desktop. Most VDI products offer features to store desktops efficiently, using linked clones or provisioning technologies. These techniques do not typically work for all desktops. They can introduce operational issues around updates and application deployments.
With SimpliVity, the underlying storage has built-in capacity efficiency. The OmniStack Data Virtualization Platform deduplicates, compresses, and optimizes data at inception, resulting in less data being transferred and stored, and eliminates redundant read/writes, improving performance. SimpliVity can store linked clones more efficiently than most storage arrays, and it can store full clones just as efficiently. SimpliVity customers can select full or linked clones based on user requirements—storage capacity does not need to be a factor.
SimpliVity provides an excellent platform for VDI. Non-disruptive hyperconverged cluster expansion allows hardware purchasing to be phased as the rollout proceeds.
By the way, SimpliVity recently published some outstanding user density numbers using the industry standard LoginVSI benchmark. You might need fewer SimpliVity nodes than you expect.