Report on a current operating systems topic

[111 home > syllabus]

Assignment

Write a 1½-page report on a current operating systems topic. Your audience for this report is a busy computer science professional who knows operating systems but may not be an expert in this particular area, and who has hired you to produce a short summary of what is important. Assume that your boss leads the information-technology R&D staff of a large aerospace corporation that develops, integrates and maintains software in both technical and financial areas specific to the company.

For this quarter’s topic, you’ll be analyzing a feature introduced in version 3.3.5 of the Hadoop distributed file system (HDFS).

Reading list

Read:

  1. Apache Software Foundation, HDFS Architecture, version 3.3.5 (2023).
  2. Apache Software Foundation, Hadoop 3.3.5 release notes (2023).
  3. Apache Software Foundation, Hadoop 3.3.5 overview of changes (2023).
  4. Apache Software Foundation, Hadoop 3.3.5 org.apache.hadoop.fs.FSDataInputStream PositionedReadble readVectored API, version 3.3.5 (2023).
  5. Mukund Thakur, Hadoop Vectored IO: your Data just got Faster! ACNA 2022. (A video of the talk is also available, but you needn’t watch it.)
  6. Linux man-pages project, preadv2(2) — manpages-dev — Debian testing — Debian Manpages, man-pages release 6.02.

What your report should do

Compare and contrast the Vectored IO API introduced in Hadoop 3.3.5 with the preadv2 system call in GNU/Linux.

  1. List important of advantages and disadvantages of Hadoop’s readVectored API compared to GNU/Linux’s preadv2.
  2. What features does readVectored have that preadv2 lacks? and vice versa?
  3. What failure modes or states does readVectored have that preadv2 lacks? and vice versa?
  4. Give an example application that should work better with readVectored than with preadv2, and vice versa. The application should be something that your boss can relate to.

Submit

Submit on Bruin Learn a 1½-page report in PDF form. Your report should be named vectored-IO.pdf ahd should follow the guidelines in Resources for written reports and oral presentations, in particular its guidelines for citing others, for USENIX templates for papers, and for how to write good papers. It is OK to go over 1½ pages if the excess material consists merely of figures, citations, and appendices.

If you prepared your report using generative AI tools like ChatGPT, your report should cite the tools you used, and should contain an appendix containing all the prompts that you gave each tool and all the responses that you received.