Michael Kasick Honored by Computing Research Association
Michael Kasick, a senior in Electrical and Computer Engineering (ECE), was selected as a finalist in the Computing Research Association's (CRA) Outstanding Undergraduate Award competition for 2007. The annual award recognizes undergraduates from North American universities who show outstanding potential in computing research.
Kasick became interested in conducting undergraduate research at Carnegie Mellon after taking Embedded Real-Time Systems, taught by Priya Narasimhan, Assistant Professor of ECE and Institute for Software Research International (ISRI). He began by volunteering with her in the summer of 2005 on a project investigating the research challenges underlying fingerpointing (also know as root-cause analysis or failure diagnosis) in large-scale distributed systems.
Later, his CIT honors project involved developing algorithms and tools to assist administrators of the real-world Emulab 400+-node cluster (located at the University of Utah) in diagnosing the root cause of failures. His results, "Towards Fingerpointing in the Emulab Dynamic Distributed System," were published last month at the USENIX Workshop on Real Large Distributed Systems (WORLDS), which was held in Seattle, in conjunction with the USENIX Symposium on Operating Systems Design and Implementation (OSDI).
"I have advised many undergraduate and graduate students, and I have to admit that Mike's prolific progress in his research has blown me away," said Narasimhan. "It is very hard for even graduate students to get their research work published at a workshop like WORLDS. Great things await Mike in the future, and I am excited and privileged to be along for the ride."
Outside of the classroom, Kasick serves in leadership roles in Carnegie Mellon's Computer Club and the Carnegie Tech Radio Club (W3VC). He has been admitted into the Ph.D. program in ECE, and plans to continue his fingerpointing research.
Students are nominated for the CRA Outstanding Undergraduate Award by their department. Other honorees from Carnegie Mellon include: winner Stephanie Rosenthal, a senior in Computer Science (CS) and Human-Computer Interaction (HCI); finalist Mihir Kedia, a senior in CS; and honorable mention nominee Brendan Meeder, a senior in CS and Math.
An announcement of the winners will appear in the January 2007 issue of Computing Research News and the awards will be presented at an upcoming computing research conference. This year's award program is sponsored by Microsoft Research.
-- Source: Computing Research Association
Carnegie Mellon Researchers Win HPC Analytics Challenge at ACM/IEEE SC2006
Computer science (CS) graduate student Tiankai Tu and David O'Hallaron, Associate Professor of CS and Electrical and Computer Engineering (ECE), led a team of researchers to win the High Performance Computing (HPC) Analytics Challenge at ACM/IEEE Supercomputing 2006 in Tampa, FL. ECE graduate student Julio Lopez was also a member of the winning group.
The entry is titled Remote Runtime Steering of Integrated Terascale Simulation and Visualization. The team developed a novel analytic capability that enables scientists and engineers to obtain insights from on-going large-scale parallel unstructured finite element mesh simulations. During the Analytics Challenge session, the team showed a live demo: steering, in real-time, the visualization of a 2050-processor earthquake ground motion simulation running on the Cray XT3 supercomputer in Pittsburgh, PA, via a wireless Internet connection, from a laptop computer in the conference room in Tampa, FL.
The Carnegie Mellon team members were Tiankai Tu (team lead), Jacobo Bielak, Julio Lopez, David O'Hallaron, Leonardo Ramirez-Guzman, and Ricardo Taborda-Rios. The other team members were: Hongfeng Yu (technical lead) and Kwan-Liu Ma of the University of California, Davis; Omar Ghattas of the University of Texas at Austin; and Nathan Stone and John Urbanic of the Pittsburgh Supercomputing Center.
-- Source: Byron Spice, Carnegie Mellon Computer Science News
Researchers Tackle Problem of Data Storage for Next-Generation Supercomputers: Carnegie Mellon leads DOE-sponsored Petascale Data Storage Institute
The U.S. Department of Energy (DOE) has awarded a five-year, $11 million grant to researchers at three universities and five national laboratories to find new ways of managing the torrent of data that will be produced by the coming generation of supercomputers. The Petascale Data Storage Institute (PDSI) combines the talents of computer scientists at Carnegie Mellon University, the University of California at Santa Cruz and the University of Michigan with those of researchers at the DOE’s Los Alamos, Sandia, Oak Ridge, Lawrence Berkeley and Pacific Northwest national laboratories.
The innovations developed by the PDSI will enable scientists to fully exploit the power of computing systems capable of performing millions of billions of calculations each second. Increased computational power is necessary to model and simulate extremely complicated phenomena, e.g. global warming, earthquake motions, the design of fuel-efficient engines, etc., which provides scientific insights into processes that are often impossible to achieve through conventional observation or experimentation.
But, simply building computers with faster processing speeds — the new target threshold is a quadrillion (a million billion) calculations per second, or a “petaflop” — will not be sufficient to achieve those goals. Garth Gibson, who will lead the data storage institute, said new methods will be needed to handle the huge amounts of data that computer simulations both use and produce, along with all the challenges that managing a system with petaflop capabilities will involve.
The PDSI will focus its efforts in three areas: collecting field data about computer failure rates and application behaviors, disseminating knowledge through best practices and standards, and developing innovative system solutions for managing petascale data storage. The latter category could include “self-*” systems that use computers to manage computers.
For more information, see the article in this newsletter beginning on page X, and visit the PDSI web site at www.pdl.cmu.edu/PDSI/.
-- with info from the Carnegie Mellon Press Release Sept. 7, 2006
Carnegie Mellon Researchers Attack Rising Costs of Data Center Operations
PITTSBURGH—Carnegie Mellon University researchers today announced the creation of the Data Center Observatory (DCO), a dual-purpose facility that is both a working data center and a research vehicle for the study of data center automation and efficiency. The university held a DCO lab dedication today at its Collaborative Innovation Center (4720 Forbes Avenue) that included representatives from the university, American Power Conversion (APC), local business leaders and select members of the news media.
The DCO is a large-scale collaborative effort between Carnegie Mellon's College of Engineering and School of Computer Science. It also includes participation from a number of industry and government partners, including APC, which is providing engineering expertise and its InfraStruXure® system for powering, cooling, racking and managing equipment in the DCO.
The DCO's principle research goals are to better comprehend and mitigate human administration costs and complexities, power and cooling challenges, and failures and their consequences. It also aims to understand resource utilization patterns and opportunities to reduce costs by sharing resources among users.
Energy efficiency is one of the center's major thrusts. For some time now, the amount of power consumed by commodity servers has been increasing, as has the number of servers placed in a facility.
"These large clusters of power-hungry machines, along with rising energy prices, are generating huge energy bills, forcing data center owners nationwide to seek more energy-efficient solutions," said Greg Ganger, a professor of electrical and computer engineering and director of the Parallel Data Lab (PDL), a Carnegie Mellon organization specializing in the study of storage systems. To tackle these issues, university researchers are working with APC to develop new ways to reduce energy demands in data centers.
Administration costs are another major research thrust. Data centers are complex to operate and require significant human administration support.
"Anecdotally, we know that human costs are a dominant part of the total cost of ownership for data centers, but exactly where people spend their time isn't well understood. One of the things that makes the DCO so interesting is that, for the first time, university researchers will be able to study human costs and efficiencies in a working data center," said Bill Courtright, executive director of the PDL.
"APC is extremely delighted to partner with Carnegie Mellon, one of the finest institutions in the world for computer engineering education and research, on this important data center initiative," said Dwight Sperry, APC vice president of Enterprise Systems and Business Networks. "We look forward to working with Carnegie Mellon to help them solve the many challenges of designing and deploying high-density data centers in the future. The Data Center Observatory faces many of the challenges common to all data center planners, such as space constraints, cooling high-density systems and the unpredictability of future growth. APC's InfraStruXure offers a space-saving, scalable, redundant, and pay-as-you-grow modular design that addresses all these concerns and delivers it at a lower total cost of ownership compared to legacy systems."
The 2,000-square-foot DCO has the ability to support 40 racks of computers, which would consume energy at a rate of up to 774 kW — more than the rate of consumption of 750 average-sized homes. In addition to studying dense computing environments, the DCO will support a variety of Carnegie Mellon research activities, from data mining to CAD/architecture, visualization and real networked services. The DCO joins Carnegie Mellon's long tradition of weaving infrastructure research into campus life, which keeps the university at the forefront of technology.
-- Carnegie Mellon University Press Release May 23, 2006
Brandon Salmon Awarded Intel Foundation Ph.D. Fellowship
Congratulations to Brandon on being awarded an Intel Foundation Ph.D. Fellowship. The Intel Foundation Ph.D. Fellowship Program awards two-year fellowships to Ph.D. candidates pursuing leading-edge work in fields related to Intel's business and research interests. Fellowships are available at select U.S. universities, by invitation only, and focus on Ph.D. students who have completed at least one year of study.
Best Demo at ICDE 2006
Congratulations to the Staged Database Systems team, who have received the Best Demo Award at ICDE 2006 in Atlanta, Georgia. Debabrata Dash, Kun Gao, Nikos Hardavellas, Stavros Harizopoulos, Ryan Johnson, Naju Mancheril, Ippokratis Pandis, Vladislav Shkapenyuk, and Anastassia Ailamaki were the collaborators on this effort titled Simultaneous Pipelining in QPipe: Exploiting Work Sharing Opportunities Across Queries. The demo paper can be found in the proceedings of the 22nd IEEE International Conference on Data Engineering (ICDE2006).
Adrian Perrig Keynote Speaker at IPSN
Adrian Perrig, assistant professor of electrical and computer engineering, engineering and public policy, and computer science, has been invited to deliver the keynote presentation at the International Conference on Information Processing in Sensor Networks (IPSN) this April in Nashville, where he will address his vision for security in sensor networks. IPSN is one of the two top conferences on sensor networks.
Adrian Perrig Recipient of 2006 Sloan Award
Three CMU 2006 winners of a Sloan Research Fellowship in computer science have been announced: Carlos Guestrin, CALD and CSD, Doug James, CSD and RI, and Adrian Perrig, ECE and CSD. A Sloan Fellowship is a prestigious award intended to enhance the careers of the very best young faculty members in specified fields of science. Currently a total of 116 fellowships are awarded annually in seven fields: chemistry, computational and evolutionary molecular biology, computer science, economics, mathematics, neuroscience, and physics. Only 14 are given in computer science each year so CMU once again shines. Congratulations to all!
Microsoft Fellowship Helps Develop New Security Course
Faculty members Lorrie Cranor, Institute for Software Research International, Jason Hong, Human-Computer Interaction Institute, and Michael Reiter, Electrical and Computer Engineering, have received a 2005 Microsoft Research Trustworthy Computing Curriculum Award to fund the development of a new course on usable privacy and security. The course is offered for the first time this semester in the School of Computer Science (http://cups.cs.cmu.edu/courses/ups.html). It is designed to introduce students to a variety of usability and user- interface problems related to privacy and security and give them experience in designing studies aimed at helping to evaluate usability issues in security and privacy systems.
-- CMU 8 1/2 x 11 News, Feb. 9, 2006 Vol. 16, No. 29
Jure Leskovec awarded Microsoft Research Fellowship
Congratulations to Jure Leskovec (PDL, CALD), who has been selected as Microsoft Research
Fellows for the next two years. Jure works with Christos Faloutsos, and is interested in link analysis and large graph mining.
The competition for these fellowships was extremely high. Their awards will be two of only 10 specially funded MSR fellowships to be funded by "a major new Microsoft initiative that will be publicly announced in the coming weeks."
James Newsome awarded Microsoft Research Fellowship
Congratulations to James for being awarded a Microsoft Research Fellowship! He won this award for his outstanding thesis work on "Sting: an automatic self-healing defense system against zero-day exploit attacks".
As stated by MSR, "[The] selection for this award is a tremendous honor and recognition of [the recipient's] accomplishments." This fellowship is one of the most prestigious fellowships for a PhD student, where each school is usually only allowed to submit up to three of their top candidates, and only less than 15% of these highly qualified candidates will be selected for the award.
PDL Researchers Recieve Both Best Paper Awards at FAST 2005!
Researchers from Carnegie Mellon's Parallel Data Lab (PDL) received both Best Paper awards at the recent File and Storage Technologies (FAST) conference, the top forum for storage systems research.
The first paper, "Ursa Minor: Versatile Cluster-based Storage," describes initial steps toward PDL's long-term target of storage systems that manage themselves (Self-* Storage). "On Multidimensional Data and Modern Disks" introduces a new approach to exploiting modern disk characteristics for better system performance. That research arises from collaboration between the PDL, Intel Research Pittsburgh, and EMC Corporation.
One of academia's premier storage systems research centers, the PDL
is an interdisciplinary group, bringing together graduate students and
faculty mainly from the Computer Science and ECE departments. Both of
the winning papers have authors from both departments.
-- ECE News Online
More PDL news here.