From: Honghai Liu (liu789_at_hotmail.com)
Date: Mon Jan 19 2004 - 13:49:20 PST
Reviewer: Honghai Liu
The paper presents an alternative architecture of kernel and applications as opposed to the one most of the most commercial operating systems are built on.
The motivation of Exokernel ( and Aegis) is that protection and management are orthogonal but kernel of most of OSs provide both, which unnecessarily limit the ability for applications to manage their own resources. Therefore, on one hand, the applications do not have any control over resource at enough low-level and it is very difficult to optimize their performance. On the other hand, the kernel provides a universal algorithm for paging, segmentations and file management, which is difficult to suit performance needs of wide range of applications.
Exokernel solution makes the kernel protect pages and disk blocks, but applications manage them. It is pointed out that the performance of the applications under Exokernel is equivalent or much better than those under popular OSs.
More interestingly, at least to me, is that Exokernel architecture makes the debugging process makes debugging process much more easier. Programming in most real-time system requires many kernel level programming, but it is almost impossible to directly debugging in kernel. If the kernel were minimized so that the many of the system calls can be moved into procedure calls, the time to develop a program could have been much more shortened substantially.
This archive was generated by hypermail 2.1.6 : Mon Jan 19 2004 - 13:49:40 PST