Assignment 3: Literature Review
Friday 15th May 23:59
You can do it alone for all 10 marks or in a group of two for a maximum of 8 marks. This is to encourage enthusiastic researchers to do it alone and the rest to pair up and still put in a good effort. No request for a group of three will be allowed.
There will no extension of deadline. The assignment requires some time. Please start early.
Absolutely no copy-paste or paraphrasing without an explicit reference and quotation marks. All plagiarism cases will be forwarded to Diciplinary Committee.
Write a 4-page research paper in ACM format using LaTeX (no other format).
You are to pick a topic and find five papers on the topic in top conferences (Best papers in OSDI, SOSP, ASPLOS are good starting points) from 2000 onwards. Both topic and papers are to be approved from me by 30th April. You are to understand the papers in depth (a viva will be conducted) and write the 4-page paper.
The paper has to contain:
- Abstract: Why problem is your topic about. How is that problem solved by the papers you discussed. What is your take from the different solutions.
- Introduction: Background and problem in your words. Put everything in context.
- Comparative Analysis: Pick aspects of your problem, make them sub-sections, and discuss how different solutions handle that aspect. This is important to not pick papers and summarize them one by one and instead compare them on different aspects. For example, free space management is an aspect of file systems and different approaches handle it in different ways. Similarly, integrity and versioning are apects whereas indexed file system and log structured file system are complete solutions. This will be the longest section of the paper with at least 2 pages.
- Conclusions: You describe here what are next issues that can be addressed in your opinion.
- References: You list here in ACM format the key five papers you base your research on and in addition any other papers you read or took something from.
Some sample areas are:
- Deduplication (We studied one way in Venti)
- Transactional Memory (An alternate to locking)
- Memory Consistency (Relaxed consistency and related issues)
- OS Verification (Race detection, Model Checking)
Looking through papers with best paper awards in recent OSDI, SOSP, and ASPLOS is a good starting point for other ideas. Topics about distributed systems are okay but you need to have the prerequisite knowledge.
Do not take this as an easy assignment. The last time I gave this assignment it resulted in the lowest average of all assignments. The main reason was that people just summarized different papers without a deeper understanding and without any comparative analysis. If you start early, you can always get feedback from me or the TA. Also, don’t keep learning LaTeX for the last day. It takes some time to get used to.