| |
An Adaptive Dynamic Request Scheduling Model for Multi-socket, Multi-core Web Servers
文摘
To implement a high-performance web server, it is necessary to fully exploit the performance of multi-core CPUs in the web server. Traditional dynamic request scheduling algorithms, such as first-come-first-served (FCFS), adopted in multi-socket, multi-core web servers, often cause multi-replica problem and ping-pong effect. We proposed a new dynamic request scheduling model for multi-socket, multi-core web servers in this paper to solve the above-mentioned issues. To fully exploit the performance of multi-socket, multi-core systems, the new model tries to allocate the dynamic requests of the same kind to the same core. Furthermore, to avoid load imbalance between the cores, the new model dynamically allocates dynamic request based on its service time and the load status of each core. We developed a simulation web server based on the new model and did some simulation experiments on it. The results of the experiments show that the new model could reduce the response time of dynamic requests in contrast to other scheduling algorithms, such as FCFS algorithm, which means the new model provides a solution to the multi-replica problem and ping-pong effect in multi-socket, multi-core systems.
| |
NGLC 2004-2010.National Geological Library of China All Rights Reserved.
Add:29 Xueyuan Rd,Haidian District,Beijing,PRC. Mail Add: 8324 mailbox 100083
For exchange or info please contact us via email.
| |