Quanta and CSAIL Explore the Cloud
Photo: Jason Dorfman, CSAIL photographer
September 20, 2011
Thanks to Quanta Computer, Inc., CSAIL researchers now have access to a mini cloud, a significant contribution to the lab’s research infrastructure.
According to Dr. Ted Chang, Quanta’s CTO, the mini cloud donation is intended to support cloud computing and academic research, and to serve as a “platform to support the T-Party project,” CSAIL’s collaborative effort with Quanta to build an integrated virtual computing environment.
The mini cloud will double the number of physical cores in CSAIL’s entire Condor Cluster, while also providing a client facing private cloud model so that lab members can create virtual machines on demand. In addition to greatly expanding the lab’s available storage and computing power, the mini cloud will allow researchers to advance groundbreaking research in a wide variety of areas.
Lab members are also looking to use the cloud as a platform for powering academic research using the cloud, according to CSAIL Assistant Director of Infrastructure Jack Costanza. Different areas where CSAIL researchers hope to apply Quanta’s cloud technology include robotics, computer vision and graphics, human-computer interaction, medical applications and computer architectures.
According to Chang, Quanta hopes to see two types of research evolve from use of the mini cloud, research within the cloud and application building on top of the cloud. CSAIL is working in conjunction with Quanta Research Cambridge (QRC) to experiment with the new mini cloud technology. QRC, a new Quanta research lab in Kendall Square, serves as a bridge between Quanta Computer, Quanta Research Institute and MIT, collaborating with faculty and students on a wide variety of research projects and searching for potential industry applications.
In addition to collaborating with CSAIL, QRC also duplicates CSAIL’s mini cloud computing efforts, from the operational set-up stages to research applications both in and on the cloud. As CSAIL researchers begin to delve deeper into the cloud, QRC will be following along closely, performing the same tasks and computations, and searching for ways to integrate research developments into Quanta products.
Will Walker, a member of QRC, currently spends at least one day a week at CSAIL working with CSAIL’s Spoken Language Systems Group (SLS), which focuses on creating technology that allows natural spoken language interaction between humans and computers. SLS is currently using the cloud to produce massive amounts of data, in particular audio-to-English processing for its WAMI (Web-Accessible Multimodal Applications) project. HearIt, one particular application now running on the cloud, allows Facebook users to search, tag and “like” photos, using a speech recognition system.
In addition to working with the group to get its programs up and running on the cloud, Walker is also bringing WAMI into the QRC cloud and working on making the application available to outside sources.
“Technology transfer works best with people not code or artifacts,” said Steve Heller, QRC Director. “This is another example of Quanta investing in people to work closely at MIT.”
Running HearIt on the cloud has allowed SLS members a better feel for how the cloud operates. Over the next few weeks, the group plans to use the cloud to run several speech-training programs, which are typically distributed over a vast number of machines to handle the extreme workload, and to create clusters of computers in the cloud.
In the future, Quanta plans to increase the computational power of CSAIL’s mini cloud, said Chang, depending on the lab’s research needs and activities.
“This is an interesting experiment for what novel research could be done if you have an unlimited amount of resources,” said Costanza. “In today’s environment where things are so time driven and data driven, having an unlimited amount of computing resources should take down any boundaries students and researchers have had in past.”
Quanta’s mini cloud collaboration with CSAIL is the foundation of an effort to kick-start a new global research network, which the company hopes will enable each student to have his or her own cloud. Chang believes that cloud computing, which offers the promise of infinite computing resources, will open up a vast new landscape of research opportunities for students and researchers worldwide.
“We are sponsoring mini cloud research at CSAIL to not only support research activities between Quanta and MIT, but also to support global research,” said Chang.
CSAIL researchers are already at work on Quanta’s mission to create a far-reaching, cloud-based information network.
“One thing we want to do is there are other groups at CSAIL that use our (speech) recognizers and we want to set up a recognizer server on the cloud so that they can just go in and use it,” explained Scott Cyphers, a research scientist with SLS.
CSAIL researchers working on the Envoy autonomous robotic assistant project, which is being developed to escort and interact with visitors and lab members, plan to tap into SLS’s speech recognizers running on the cloud to enable speech recognition services on the robotic assistant.
"Using this in-house cloud, we'll be able to experiment with many different cloud software architectures. There are some exciting opportunities in scalable computing made possible by moving beyond the simple virtual machine interface offered by commercial cloud offerings,” explained Chris Terman, T-Party research director and co-director of CSAIL. “Our work in areas such as operating systems for multi-core clusters and relational databases-as-a-service will take advantage of the accessibility afforded by Quanta's generous equipment donation."
Abby Abazorius, CSAIL