Application of collaborative process Group memory management support technology
Chen Baozi Wu Qingbo Tan Yusung
One of the most important ways for cloud computing to aggregate resources is to combine the applications of different users and different features into a hybrid deployment and run simultaneously. In contrast, the user-state application of the garbage collector for the service individual memory management is more targeted, and the operating system for the overall memory resource allocation ability is stronger. The mechanism of the existing kernel only ensures that the service is passively garbage collected when the global memory or process group memory usage is up to the limit. Combining the Process Control group mechanism in Linux kernel and the EVENTFD event notification mechanism, a simple and efficient kernel support mechanism for the application of collaborative packet memory management is designed and implemented. By adding the memory management mechanism in the kernel, the system can support the application of autonomic memory. The experiment shows that the new mechanism has no obvious performance effect on the original operating system.
Application of collaborative process Group memory management support technology