The problems you're having may not be the problems you think you're having: results from a latency study of Windows NT

This paper is intended to catalyze discussions on two intertwined systems topics. First, it presents early results from a latency study of Windows NT that identifies some specific causes of long thread scheduling latencies, many of which delay the dispatching of runnable threads for tens of milliseconds. Reasons for these delays, including technical, methodological, and economic are presented and possible solutions are discussed. Secondly, and equally importantly, it is intended to serve as a cautionary tale against believing one's own intuition about the causes of poor system performance. We went into this study believing we understood a number of the causes of these delays, with our beliefs informed more by conventional wisdom and hunches than data. In nearly all cases the reasons we discovered via instrumentation and measurement surprised us. In fact, some directly contradicted "facts" we thought we "knew".

[1]  Jason Nieh,et al.  The design, implementation and evaluation of SMART: a scheduler for multimedia applications , 1997, SOSP.

[2]  Michael B. Jones,et al.  CPU reservations and time constraints: efficient, predictable scheduling of independent activities , 1997, SOSP.

[3]  William J. Bolosky,et al.  Distributed schedule management in the Tiger video fileserver , 1997, SOSP.

[4]  Lui Sha,et al.  Priority Inheritance Protocols: An Approach to Real-Time Synchronization , 1990, IEEE Trans. Computers.

[5]  Peter Druschel,et al.  Resource containers: a new facility for resource management in server systems , 1999, OSDI '99.

[6]  David A. Solomon,et al.  Inside windows nt second edition , 1998 .

[7]  Krithi Ramamritham,et al.  Using Windows NT for Real-Time Applications: Experimental Observations and Recommendations , 1998, IEEE Real Time Technology and Applications Symposium.

[8]  Michael B. Jones,et al.  An overview of the Rialto real-time architecture , 1996, EW 7.

[9]  J. C. Mogul.  SPECmarks are leading us astray , 1992, [1992] Proceedings Third Workshop on Workstation Operating Systems.

[10]  Zheng Wang,et al.  Using latency to evaluate interactive system performance , 1996, OSDI '96.

[11]  J. Duane Northcutt,et al.  SVR4UNIX Scheduler Unacceptable for Multimedia Applications , 1993, NOSSDAV.

[12]  Stefan Savage,et al.  Processor capacity reserves: operating system support for multimedia applications , 1994, 1994 Proceedings of IEEE International Conference on Multimedia Computing and Systems.