It’ not what VMware wants to hear, but I/O is one of the big bottlenecks in virtualized server environments.

But one vendor’s challenge is another vendor’s opportunity, according to Stephen Spellicy.

Spellicy is vice president of marketing and business development at Virtensys. The start-up, which was founded by Xiotech alumni, focuses on I/O virtualization. Speaking live inside theCube from VMworld 2011 in Las Vegas, Spellicy explained to Dave Vellante, Wikibon’s chief analyst, and John Furrier, founder of SiliconANGLE.com, just how Virtensys tackles the I/O problem in virtualized environments.

Said Spellicy: “At core what we do is take standard, off-the-shelf adapters in a client-style solution, we plug those into our chassis and then we share, using PCI Express-based switching and a virtual proxy controller, all of the virtualization layer of the hardware that’s being inserted in our device. The technology is sold for standard server environments, standard rack environments. And we also have an OEM arrangement with NEC Corporation in Japan and they’re selling a blade-version of our product.”

Spellicy said virtualized I/O provides systems administrators both greater choice and control of the types of interfaces that they present to standard servers.  They can drive these servers harder because they’re offloading the I/O process to the hardware.

“So actually you’re taking a lot of the overhead and burden off the traditional host, lower CPU utilization on the server host, greater performance, and your I/O performance is increased because you’re driving both the resources and that appliance or infrastructure harder and you’re getting more out of it through acceleration.”

Speaking about cloud computing generally, Spellicy told Vellante and Furrier that the key to adoption is making the systems administrator’s job as simple as can be and making it possible to meet varying levels of SLAs. Virtensys helps customers do both via Quality-of-Service and bandwidth allocation capabilities.

“An admin can literally go in and fine-tune the percentage of guaranteed bandwidth on a given interface. That virtualized 10Gb pipe can be assigned to any server in your pod and then you can literally govern, monitor and watch how that’s being utilized by customer or by environment.”