Abstract
Main memory management is critical for enhancing the performance of web server systems that include biological information. For decreasing the transaction time of incoming requests from users, such systems create several processes for future requests, saving the time needed to create processes that handle incoming requests from users. However, inefficient process management can decrease the performance of web server systems, and web cache systems connecting through proxy servers create dynamic access patterns that make it difficult to predict how many requests are coming into a system. Furthermore, while persistent and pipeline schemes decrease transaction time of incoming requests by sending multiple requests at the same time, these schemes waste available memory space by requiring multiple processes in order to handle multiple connections. Too many active processes result in a reduction of the system's overall performance. Therefore, we suggest an adaptive process management scheme through web log mining. In our scheme, the numbers of web processes are controlled through prediction of incoming requests. Our management of processes saves on available memory without decreasing transaction time. We also demonstrate the effectiveness of our scheme through application to real web workload.
Original language | English |
---|---|
Article number | 104 |
Pages (from-to) | 716-720 |
Number of pages | 5 |
Journal | Life Science Journal |
Volume | 11 |
Issue number | 7 |
State | Published - 2014 |
Keywords
- Biological web pipeline
- Web log mining
- Web server systems