ICAR Research Data Repository for Knowledge Management
DSpace at IIT Bombay
View Archive InfoMetadata
| Field | Value |
| Title | An online, derivative-free optimization approach to auto-tuning of computing systems |
| Names |
POOJARY, S
RAGHAVENDRA, R MANJUNATH, D |
| Date Issued | 2010 (iso8601) |
| Abstract | We develop an online optimisation framework for self tuning of computer systems. Towards this, we first discuss suitable objective functions. We then develop an iterative technique that is robust to noisy measurements of objective function and also requires fewer perturbations on the configuration. We essentially adapt the Nelder-Mead algorithm to work with constrained variables and also allow noisy measurements. Extensive experimental results on a queueing model and on an actual system illustrate the performance of our scheme. |
| Genre | Proceedings Paper |
| Topic | Server |
| Identifier | DISTRIBUTED COMPUTING AND NETWORKING, PROCEEDINGS,5935,434-445 |