Virtualization has taken the IT world by storm, and it's now readily accepted as the go-to technology for organizations looking to wring more productivity out of underutilized hardware and to take a huge bite out of IT infrastructure budgets.
Yet while the technology tops the must-have lists of most CIOs, it hasn't captured nearly the same attention in engineering circles as a means of doing more with less.
Much of why virtualization hasn't become the same dominant force for engineering systems is due to latency and performance concerns when running graphics-rich applications like CAD and CAE in this type of environment. Parallels, which specializes in virtualization and automation software, has come out with a new platform that employs the latest Intel direct assignment technology to address this problem and make virtualization a much more viable option for engineering applications running on high-end workstations.
The new Parallels Workstation 6 Extreme solution utilizes the intelligent performance features of the Intel 5500 and 5600 series chipsets, including support for Intel Virtualization Technology for Directed I/O. VT-d, or direct assignment as it's called, is technology that changes the way I/O is directed to target devices so virtual machines can interact with dedicated resources without incurring any degradation in performance that is typically associated with hypervisor emulation layers.
As result of the Intel VT-d support, graphics-intensive programs like CAD and CAE can be directly assigned to interact with graphics cards like those from Nvidia, as well as CPU and memory resources, thus bolstering their ability to run at near-native speeds, according to company claims and reports by users.
With the performance issue out of the way, Parallels can offer a pretty compelling virtualization story for high-end workstation users in a couple of different scenarios.
In one type of setup, Parallels Workstation 6 Extreme can transform a group of workstations into a distributed computing cluster, yet still protect the workstation's individual resources so an engineering team might run a computer-intensive simulation job on the cluster at the same time a single engineer models parts in CAD. In another scenario, an engineer could run a Linux-based simulation program along with a high-end, graphics-rendering application on the same workstation simultaneously, without having to switch off between dueling machines or deal with a single system with sacrificed performance.