![]() |
ANU College of Engineering and Computer Science
School of Computer Science
|
|
|
Computer NetworksBuffer BloatBuffer Bloat refers to a phenonemon that has been observed on the Internet for most of the past decade. The term "bufferbloat" was coined by Jim Gettys in December 2010. You are required to conduct your own research on the following questions and provide approximately 2 - 3 paragraphs (half a printed A4 page) per answer. You are required to submit your answers as a plain text file, or as a portable document format (pdf) file. This assignment is worth 10% of the final assessment for your respective Computer Networking course. For this assignment, half the marks (5/10) will be awarded for the quality and diversity of your citations. More marks will be awarded for citing of work that you have discovered yourself (eg. not from wikipedia). Citing of academically refereed works (eg. journal articles) will attract the most marks. If citing from a community-based reference website such as wikipedia, include the time and date that the information you are referencing was obtained from the source. Question 1: Clearly explain what is meant by "bufferbloat" and what the main observable symptoms are. Question 2: Explain the base assumptions used in the original design of TCP that exacerbate the buffer bloat phenonemon. Question 3: Clearly explain at least one technique for detecting and measuring buffer bloat. Question 4: Explain two mechanisms for mitigating against buffer bloat. Include at least one AQM mechanism such as RED or one of it's derivatives. Question 5: The "rsync" program was developed at ANU for remotely synchronising large amounts of data. Explain the command line option that can be used to allow rsync operations to continue "in the background" without adversely contributing to buffer bloat. This is an example of application-layer mitigation of the buffer bloat problem.
Extra Details for Masters Students (COMP6331/ENGN6535)Masters level students need to complete the following question, worth 4 marks including citations. Your assignments will be marked out of 14 and scaled to be worth 10%. Masters level additional question: Examine the implications of the bandwidth-delay product of 100 Gbps "long-haul" (greater than 200km) network links on buffer sizing and the resultant issues for buffer bloat. Provide some numerical analysis with reasonable assumptions. Submission DetailsYour submission should include references to all material used in formulating your responses (see above).
submit comp3310 Assign1 assign1.pdf or submit comp3310 Assign1 assign1.txt | |||||||||||||||||||||||||||||||||||||||||||
|
Please direct all enquiries to: bob@cs.anu.edu.au Page authorised by: Head of School, SoCS |
| The Australian National University — CRICOS Provider Number 00120C |