Award Finalist/Winner
Student Contribution

SC Conference - Activity Details

Characterizing Application Sensitivity to OS Interference using Kernel-Level Noise Injection

Kurt B. Ferreira  (University of New Mexico)
Ron Brightwell  (Sandia National Laboratories)
Patrick Bridges  (University of New Mexico)
Papers Session
Runtime Systems
Tuesday,  03:30PM - 04:00PM
Room Ballroom F
Operating system noise has been identified as a key limiter in application scalability. Several studies have attempted to quantify the sources and effects of noise using user-level mechanisms. In this study, we describe a kernel-based mechanism for noise injection and focus on the sensitivity of applications to various noise patterns. We describe the implementation of our noise injection framework in the Catamount lightweight kernel and show the impact of various types of OS-generated interference on applications at scales up to ten thousand nodes. Our results characterize the importance of how noise is generated, in terms of frequency and duration, and how the impact changes with scale. For example, results show that 2.5% net processor noise at 10,000 nodes can have no impact or can result in a factor of 20 slowdown, depending solely on how the noise is generated.
The full paper can be found in the IEEE Xplore Digital Library and ACM Digital Library
   IEEE Computer Society  /  ACM     2 0   Y E A R S   -   U N L E A S H I N G   T H E   P O W E R   O F   H P C