5 Key Considerations for a Successful VDI Deployment
With IT resources constantly being strained, organizations are looking for ways to gain cost efficiencies while increasing flexibility and ease of management. Enter the virtual desktop infrastructure (VDI) where desktop operating systems are hosted within virtual machines on a centralized server. This market is expected to grow at a CAGR of 29.7% from 2014-2019 across all major industry verticals. Clearly, VDI has the attention of CIOs – but not all virtualization is the same. VDI has unique considerations that must be addressed in order to achieve the business and IT objectives that the solution promises. Unfortunately, because of the highly-desired outcomes of a VDI solution, oftentimes we can miss planning for certain pitfalls when architecting a remote desktop solution.
Below are five common factors that are frequently overlooked when designing VDI:
1. User experience
VDI is unlike typical client-server applications due to the nearly real-time expectation from users. There is little tolerance for perceptible delays from a keystroke, or mouse movements—nor should there be. Users are accustomed to local desktop performance so any perceptible reduction in response time often jeopardizes the entire project ; in fact 59% of those polled in a survey indicated that end user experience as the biggest challenge after deploying VDI.
Typical VDI solutions are deployed on centralized infrastructure in the data center. Since LAN-like desktop speeds are the expectation, the solution often falls short leaving users highly frustrated. IT must consider an architecture where desktop computing is closer to the end user regardless of location.
2. User productivity
The two most common factors that affect user productivity in VDI deployments are WAN performance and WAN outages. Since computing resources are consolidated in the data center, a typical deployment is directly dependent on an oversized WAN link back to the data center. Variations in network bandwidth, latency, time of day, and number of simultaneous users logging in, all can adversely impact productivity.
Consolidation typically benefits IT—due to centralized management and provisioning. However, it can handcuff the line of business user who demands ‘always on’ access regardless of the wide area network conditions. A successful VDI architecture must have fewer network and infrastructure dependencies to avoid loss of productivity.
3. Hidden costs
Two of the most common underestimated costs are WAN and backend storage. When adding a VDI service, additional load will be placed on the network. VDI can heavily tax a WAN circuit depending on number of users and type of applications that users are running. Often bandwidth needs to be added in order to accommodate these demands. From a data storage perspective, additional I/O load and capacity will be placed on the storage subsystem to support multiple VDI sessions. In fact, it’s estimated that for every $1 spent on VDI, $3-$10 are spent on storage.
When designing VDI, IT organizations should look for solutions where they can use a lower tier back-end storage, while simultaneously the least amount of load is added to the network to avoid unnecessary expenditures.
4. Local services
While VDI brings clear operational benefits by centralizing desktops to the data center and moving them closer to applications, local services at branch locations are often overlooked. 77% of enterprises require access to file, print and other critical, line of business services at the branch.
As majority of the branches have local services, architectural considerations need to be made to deal with the interactions between the users and local applications.
5. Over provisioning
Most VDI solutions are over-provisioned and underutilized. Both WAN connections and backend storage must be built to withstand the worst-case scenarios e.g., bootstorms and large service pack updates. As such, storage and network architects often end up allocating too much capacity to handle these one off situations.
A VDI solution must absorb these sudden network and storage spikes without over-engineering the data center and ROBO infrastructure.
The five commonly overlooked factors point to an architectural shift from a fully consolidated (at the data center) to a parity architecture, with infrastructure at both the data center and the remote office/branch office. Don’t fret! There is hope for those struggling with this delicate balancing act. Stay tuned for the next blog, where I’ll discuss how Riverbed addresses these challenges with a purpose built Hyper-Converged Edge solution for VDI.