Memory Resource Management in VMware ESX Server 문서 원본 보기

[[분류: 시스템 논문]]

 OSDI 2022 DOI: 10.1145/844128.844146

Paper review: PAPER TITLE

1. Summarize the problem(s) the paper want to solve.
Slow performance and large foot print of memory resources at full virtualization.

2. Why is/are the problem(s) important? State your opinion.
Memory management is the one the most important and sophisticated part in virtualization techniques. Thus, reduce foot print and increase utilization of guest operating system’s memory resources is critical factor in VM. In this paper, they though how to do such works in traditional full virtualization without modifying guest OS (transparently). 

3  Summarize the (at most) key main ideas of the paper.
Transparent Page Sharing: Share different’s OS pages with same host pages by hasing. 
Ballon driver: Optimizing guest OS memory space by hypervisor without modifying underlying kernel sources. 
Memory polices: Idle memory tax.

4. State the main contributions of the paper.
This paper present the mechanisms and policies of optimizing type 1 hypervisor in virtualization. This paper’s main idea is not only fresh but also impacting many other works by it’s novelty and importances. I surprised that what I learn in previous graduated school about ballon driver and memory optimization in virtualization is all originated in this paper. Also, the introduction is well organized that I could learn lots of virtualization memory management technique such as memory reclamation or overcommit. 

5. Criticize the main contributions (Limitations of the proposed design. Limitations of applicability etc.)
Rate the significance of the paper on a scale of 5 (breakthrough), 4 (significant contribution), 3 (modest contribution), 
2 (incremental contribution), 1 (no contribution or negative contribution). Explain your rating.
This paper’s idea has novelty and impaction to other researches. Lots of afterward virtualization VMM such as XEN, still use ballon driver or memory overcommit and sharing for their own purposes. Also I believed that this paper is good starting point of what is happening inside the VMM’s memory management policy and mechanisms. Thus I will give 5 to this paper.

6. Rate how convincing is the evaluation methodology (Refer to the following questions).
Do the claims and conclusions follow from the experiments? 
Are the assumptions realistic? Are the experiments well designed? 
Are there different experiments that would be more convincing? 
Are there other alternatives the authors should have considered? (And, of course, is the paper free of methodological errors.)
Their paper organize evaluation slightly different from other papers. However with plenty of explanation, I think that their evaluation is easy to understand and important to analysis what’s happening inside their new (but old) hypervisor. I think the way of evolution structure and explanation is well written for proving there new hypervisor’s advantages. 

7. 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? 
What is the lasting impact of this work? 
-> Mechanism in memory overcommit and reclamation. 
What (if any) questions does this work leave open?