TORONTO – Computer science students at the University of Toronto are developing an “easy button” for cloud computing, a delete button for spam over IP telephony and way to place buttons on the back of a mobile device.
The school on Tuesday afternoon hosted Research In Action, a cross-section of exploratory IT projects being conducted by its approximately 50 faculty and 300 graduate students. Several of the projects on display were funded by well-known private sector partners, and a few have already been commercialized or are becoming the flagship offering of new startups.
A case in point is SnowFlock, a way to rapidly clone virtual machines for users in high-performance cloud computing environments. According to graduate student Philip Patchin, the open source software is based on some Python scripts and a C library using an older version of Citrix’s Xen hypervisor. The idea behind SnowFlock, he said, is to ease the management burden associated with setting up dozens of virtual machines at one time.
“If you’re doing 100 servers, say, you have to start each one up, configure it, make sure the integration is there so they can talk to each other,” he said. “It can mean something that has a lot of promise ends up being a lot of grunt work.”
SnowFlock uses the message passing interface (MPI) to rapidly clone virtual machines (VMs) as they’re needed and collapse them once they’re no longer required, Patchin said. “You could easily set up a cluster and manage all your applications across these VMs,” he said, adding the software is of particular interest to academic institutions. Funded by Rogers Communications, SnowFlock has since been spun off into a company called GridCentric, Patchin said, which has set up a small team on the Rogers campus in Toronto.
While cloud computing and virtualization are starting to infiltrate the enterprise, another group of U of T students is focused on a more traditional problem in a new form. Alireza Sahraei is one of three graduate students trying to prevent SPIT, otherwise known as spam over IP telephony. While not a major issue today, Sahraei said the team believes SPIT will escalate considerably as voice-over-IP costs come down and mobile devices continue to proliferate among consumers.
“It’s a much bigger issue than e-mail,” he said. “E-mail can be annoying, but it just sits there. If you get a (spam) call over IP when it’s 2:00 a.m., you’re going to get pretty angry.”
Sahraei’s team is working on a “central SPIT detector,” which would monitor IP traffic and send information to a central server. As spam is detected notifications would be sent to the appropriate gateways. It’s not as easy a project as it might sound.
“You need to be able to detect it in real-time,” he said, in order to prevent SPIT calls before they happen. That requires information on patterns and trends in VoIP traffic, which service providers are reluctant to provide for privacy reasons.
Perhaps the best example of out-of-the-box thinking at Research In Action came from a project called RearType, a collaboration between U of T and Microsoft Research which is exploring putting text entry keys on the back of a device. Leila Rezai, who was presenting a prototype of the team’s early work, said the idea was to free up the front of a device like a tablet PC while still providing a keyboard type of experience. The keys are placed vertically in roughly the place where a user would clutch the device with each hand, providing multi-finger typing.
“We’ve found that in one hour’s training, the entry speeds were 15 words per minute,” Rezai said, which is in keeping with typical speed on a touchscreen keyboard. The team is now looking at how RearType could be applied to smaller devices, about the size of a BlackBerry.
Other projects at Research In Action included debugging domain-specific languages and a system called Lockr to provide better privacy over social networks.
