About me
 |
I am a sixth year Ph.D student at Carnegie Mellon University, in the Electrical & Computer Engineering department. I work in the Parallel Data Lab (PDL) and am advised by Professor Greg Ganger. My research interests span the field of computer systems, including file systems, with a focus on cloud computing. In 2007, I was featured in a PhDComics strip.
I am specifically interested in building self-healing distributed services, which are capable of fixing problems automatically without human intervention. Such automation is necessary, given the rapidly increasing scale and complexity of modern clouds and datacenters.
As a first step, I have been researching techniques for helping developers diagnose complex performance problems in distributed services---for example, by automatically localizing the root cause of a problem from the myriad components in the system to just a few relevant interactions. I believe these techniques will serve as a strong base on which to build further automation. For my future research, I am interested in collaborating with researchers in machine learning, statistics, and HCI to take additional steps toward the larger goal.
Thesis work
My dissertation research focuses on a technique, called request-flow comparison, for automatically localizing the source of performance changes observed in a distributed service. These changes occur frequently in shared-machine environments, such as most datacenters, and even in more traditional dedicated-machine environments. As such, they are an important area on which to focus when building automation tools.
Request-flow comparison leverages the key insight that performance changes often manifest as mutations in the path requests take through the distributed service---e.g., the components they visit and the functions they access---or their timing. Exposing these mutations and showing how they differ from previous behaviour localizes the source of the problem and significantly guides developer effort. In our NSDI'11 paper, we showed the effectiveness of request-flow comparison by using it to diagnose real, previously undiagnosed problems in a prototype distributed storage service and in certain Google services.
Selected publications
- Diagnosing performance changes by comparing request flows. Raja R. Sambasivan, Alice X. Zheng, Michael De Rosa, Elie Krevat, Spencer Whitman, Michael Stroucken, William Wang, Lianghong Xu, Gregory R. Ganger. NSDI'11.
[Abstract] [Paper] [Talk video] - Categorizing and differencing system behaviours. Raja R. Sambasivan, Alice X. Zheng, Eno Thereska, Gregory R. Ganger. HotAC'II.
[Abstract] [Paper] - Ursa Minor: Versatile cluster-based storage. Michael Abd-El-Malek, William V. Courtright II, Chuck Cranor, Gregory R. Ganger, James Hendricks, Andrew J. Klosterman, Michael Mesnier, Manish Prasad, Brandon Salmon, Raja R. Sambasivan, Shafeeq Sinnamohideen, John D. Strunk, Eno Thereska, Matthew Wachs, Jay J. Wylie. FAST'05.
[Abstract] [Paper] - //TRACE: parallel trace replay with approximate causal events. Michael Mesnier, Matthew Wachs, Raja R. Sambasivan, Julio Lopez, James Hendricks, Gregory R. Ganger. FAST'07.
[Abstract] [Paper]