[123] in 6.033-lab
Re: 033 lab
daemon@ATHENA.MIT.EDU (Kevin Fu)
Thu Feb 25 20:55:56 1999
To: Me! <strider@MIT.EDU>
Cc: 6.033-lab@MIT.EDU
In-Reply-To: Your message of Thu, 25 Feb 1999 19:07:38 -0500.
<199902260007.TAA21763@w20-575-23>
Date: Thu, 25 Feb 1999 20:55:47 EST
From: Kevin Fu <fubob@MIT.EDU>
Hi,
[CC'ing the other students since these are important questions.]
>I am somewhat confused about the web proxy assignment. So, here are a few
>questions to create more confusion for myself:
>
>1) Caching, in its basic form, constitutes only caching in RAM, correct? So
>designing a cache with RAM / disk files would constitute an "optimization" ?
You'll find a disk cache useful since memory is small, disk space is
large, and HTTP responses consume lots of memory. A disk cache is not
required, but your proxy may fail or thrash if you implement a
memory-only cache. Of course, the pages on disk must flow through
memory. Designing a disk cache alone does not constitute an
optimization. To qualify, the proxy must have a clear multi-level
caching scheme. For example, you may swap old pages out of memory to
disk. How you cache pages is an important part of your design.
>2) When caching, is taking into consideration Expires, etc. required or an
>extra?
Yes, this is indirectly required because your proxy should correctly
handle any HTTP/1.0 GET, POST, or HEAD request (see top of page 2 of
the handout). If you search for the keyword "proxy" in RFC 1945,
you'll see all the requirements of a proxy w.r.t. HTTP/1.0.
>3) Which of the headers actually require intervention? This did not seem to be
>very clear. Is all of the relevant info in rfc1945, or are there other rfc files
>of interest?
The proxy will blindly forward most headers from the client to the
server and vice-versa, but some headers require intervention. For a
list of these important headers, see the middle of page 4 of the web
proxy handout.
RFC 1945 specifies all the headers you need to know about.
>4) Is there a particular part of the Mayfield Handbook which is of relevance for
>the write-up? Just out of curiosity...
Yes, look in subsection 2.4.4 (Design and Feasibility Reports). In
hard copy, this is page 80. This section is extremely relevant to
6.033 as you can discover by reading the sample report...
Last year's excellent reports also demonstrate good design writeups.
Select "1998 pages" from the 6.033 web page.
>5) Can we find out what the proxy is going to be tested upon? Are any of the
>proxy-related issues mentioned outside of sections related to the headers listed
>on p.4 of the handout, and outside of the sections related to GET, POST and
>HEAD, required for this web proxy instantiation?
>
>Ovidiu
Along with simple tests (using your web proxy on various pages), we
will test for the requirements given in RFC 1945 and the web proxy
handout. Things to look for:
Your web proxy should handle text pages and binary files. It should
also respect the limited resources of the client machine, proxy
machine, and server machine. Filling my disk is a bad thing. :-)
Hope that helps!
[In the future, please email 6.033-lab-tas@mit.edu rather than me
personally.]
--------
Kevin E. Fu (fubob@mit.edu)
PGP key: finger fubob@monk.mit.edu
