Manufacturing and design engineers, at Intel, are always in pursuit of faster and better personal desktop workstations, or High Performance PCs (HPPC). Unfortunately in today’s economic climate it’s difficult to justify spending thousands of dollars on a high performance workstation for each individual. Especially if the desktops may need to be upgraded in a few years’ time thus reigniting the whole purchase process all over again.
Fortunately there are other options which easily scale out, perform better and cheaper in the long run. I’m referring to a virtualized desktop in the “Cloud” (I promise this is the only time I’ll mention this already strained term). For a fraction of the TCO of a few hundred HPPCs we have setup virtual desktop farms which can support twice as many virtual desktops, on an enterprise hypervisor running on Intel’s latest virtualization optimized Xeon processor line.
The key is to keep the solution as cost effective as possible without going overboard. It’s easy to go crazy and design a tier one solution but those can quickly undermine the ROI. We were sure to only add the bare necessities and mitigate the risks with a sound business continuity plan. Remember these aren’t tier one servers, however hundreds of angry engineers can easily make a tier one like uproar if this environment were to go down for an extended period of time.
Desktop virtualization provides many other advantages which a physical workstation may not provide, regardless of its endowment. By virtualizing, we are sharing the pooled resources. Using every bit of the pool before scaling out and not wasting CPU cycles on workstations sitting idle. Allowing us to dynamically allocate resources to where they are needed and eliminate idle waste.
An enterprise capable hypervisor, running on Intel virtualization optimized hardware, can magically seem to improve response and processing time. The magic lay in the benefits from pooling. Such as caching the frequently used virtual desktop OS files to memory for all to access without incurring disk IO. Another benefit, usually seen when virtualizing single threaded applications is an improvement in overall runtime. Sure it helps to be running on a server platform however the main reason is in part thanks to the hypervisor. The hypervisor provides an abstracted virtual CPU, used by the desktop application thread and optimizes its runtime by spreading the load across multiple physical CPUs and increasing the number of CPU cycles we can run.
By virtualizing, we can now make use of more automated provisioning technologies and spend less resources maintaining and building physical workstation OS. Provisioning a desktop OS no longer requires an individual to walk up to the terminal and install an OS. We no longer need to install an OS for that fact. We applied industry best known methods for automating virtual desktop cloning and customization hence introducing yet another cost benefit and improving cycle time.desktop
Virtualization allowed us to provide more to our customers, cheaper better faster. It changes the operations mindset and opens new doors to improve the way we do our business. Makes better use of physical hardware so that we could use as much of the hardware as we are paying for with as little waste as possible.
Lichaah is a Senior Infrastructure Engineer with more than 15 years of Information Technology experience in both UNIX and Windows environments. At work, he enjoys applying his software engineering skill to architecting practical solutions.\
Check out his previous posts and discussions