ACM Transactions on Computer Systems (TOCS) 1994
Summarize the problem(s) the paper want to solve.
-> Performance issue and related problems of memory sharing in private virtual address space. -> Old-fasioned desing concept of private virtual address space in "new" 64bit high "performance" sytstem. ("_" in 1990's)
Why is/are the problem(s) important? State your opinion.
In heavy memory sharing environment such as database systems, private virtual address space and process machanism hinder the pointer-based direct memory sharing. So, persistent memory object or dynamic allocation etc, are not capable in traditional private virtual address based protections.
Summarize the (at most) key main ideas of the paper
Use single address space, make all object in memory can mutually visible in all processes (protection domain) in operating system. To provide protection like before, seperate context information and the other from UNIX like process design, and give managed access to each segments by password based capability model.
State the main contributions of the paper
-> Novel view of seperating context (protection) information from process. -> Give idea of single address space for operating design. -> Prove that even in single address space OS, protection between each protection domain can be achived.
Criticize the main contributions (Limitations of the proposed design. Limitations of applicability etc.)
-> The concept of the providing shared object by single address space is somewhat old fashioned in nowadays. Because majority of UNIX like operating system including LINUX also provide almost all advantages of SASOS. -> But, the concept of seperating protection from process is affecting lots of future generations papers. -> Also this idea is really interesting and thinkable. -> And there is lots of engineering efforts in this paper. -> So I will give 5/5.
Rate how convincing is the evaluation methodology (Refer to the following questions)
-> They design CAD system to test SASOS performance. Compare with monolitic(SASOS wihtout protection), difference is relatively small. -> Also Decomposed version and SASOS has high performance difference. -> But, their is no information of micro benchmark such as mmap's latency or throughput. -> So I think evaluation of this paper is weak but have sufficient convincing information of proving.
Answer one of the following three questions (whichever is most relevant for this paper): What lessons should system researchers and builders take away from this work?
-> Concept of seperating protection domain from traditional process. -> Single address operating system and engineering efforts.
What is the lasting impact of this work?
-> Novel design concept of SASOS.
What (if any) questions does this work leave open?
-> Implementing segment protection by hardware not in software. -> Language based Single Address Operating system such as RUST or GO.
2 - 3 sentences are fine for each question..