Server: Netscape-Commerce/1.1 Date: Tuesday, 26-Nov-96 17:58:20 GMT Last-modified: Monday, 26-Aug-96 16:07:21 GMT Content-length: 3263 Content-type: text/html
My research interests include modeling and performance evaluation of computer systems. My dissertation topic is in the area of performance modeling and prediction of parallel systems. The title of the dissertation is: Analysis of parallel systems with fork-join synchronization constraints.
My dissertation work primarily consists of two distinct papers. The first paper presents an exact response time analysis of open and closed parallel systems with service times drawn from an exponential distribution. The analysis quantifies the recursive behavior of fork-join systems and helps to better understand the effect that parallelism has on system performance. The methods used are classical queueing theory and performance evaluation techniques, and basic theory of stochastic processes. A paper on this work will be submitted to the Journal of ACM later this month. A preliminary paper on the mean response time analysis of 2-sibling fork-join systems appeared in MASCOTS'96.
The second paper presents an exact analysis of closed fork-join systems and introduces a quick bounding technique for arbitrary fork-join systems. The bounding technique is similar to balanced job bounds for product-form networks. For this paper, the methods used are earlier results on product-form networks and Markov analysis. This paper appeared in SIGMETRICS'96.
I am also working on the solution set to performance supplement textbook, Solutions to P. S. to Computer Architecture (coming soon), jointly sponsored by CMG and ACM/SIGMETRICS.
Recent papers/publications include: