README
- Requirements:
- Perl - I run simulations
with Perl for Windows but with minor changes they can be run under any
Unix-like OS. The only difference is that you have to add full path to
files which are processed under Unix and
#!/usr/local/bin/perl
(full path to Perl
interpreter) is necessary for scripts in Unix.
(e.g. filter.plwhich I run on Solaris)
- Basic knowledge of Perl
- Traces which can be obtained on address:
ftp://ftp.ircache.net/Traces/
for more information visit
http://ircache.nlanr.net/
Note: Those traces are updated daily.
- A lot of disk space for traces and a lot of
memory for simulation
- Installation:
- Download traces for number of days you want to
simulate
- Unzip
them in the same directory and add extension .trc
- Run
filter.pl to make one very large file. This program filters traces and
put only necessary data for every URL in that file. It also renames URLs
with numbers to reduce amount of data to be stored in the operating
memory during simulation.
- There
are few algorithms simulated fifo (fifo.pl), lru (lru.pl), lfd (lfd.pl), infinite (infinite.pl) and pd (pd.pl) (the one I'm currently working on)
- Run
- under
Windows type
perl filename.pl
or under Unix-like OS type
perl ./filename.pl
- for
some programs it is necessary to manually create log file. Create simple
and empty text file with required file name
- Report
- Results of a simulation are reported on the screen and in the log
file. Each algorithm has it's own log file and results of following
simulations are appended to previous results.
- hit
rate and weighted (byte) hit rate are reported
- expiration date is not accounted
- Notes:
- Any bugs and questions forward to dwecci@sezampro.yu .
- Other files in the current directory are not
necessary for simulation and therefore are not explained. If you have
questions about them forward questions to dwecci@sezampro.yu .
- Have a nice simulation :) .