readme

柯老师网站上找到的

# $Header: /nfs/jade/vint/CVSROOT/ns-2/indep-utils/webtrace-conv/README,v 1.1 1999/02/24 01:19:52 haoboy Exp $

1. Traces
---------

This directory contains parsers to convert 4 types of proxy traces to input 
files as required by PagePool/ProxyTrace.

ucb/: UCB Home-IP trace, available at 
  The code is partially taken from Steven Gribble's 
  original trace parser. 

dec/: DEC proxy trace, available at 
  ftp://ftp.digital.com/pub/DEC/traces/proxy/webtraces.html. The code is 
  partially taken from Alex Rousskov's trace parser (available at 
  ftp://ftp.digital.com/pub/DEC/traces/proxy/contrib/proxytrace2any.tar.gz).

epa/: EPA server trace, available at
  http://ita.ee.lbl.gov/html/contrib/EPA-HTTP.html.

nlanr/: NLANR proxy trace, refreshed daily and available at:
  ftp://ircache.nlanr.net/Traces/


2. Usage
--------

In any of these directories, doing a make will produce an executable 
'tr-stat'. It assigns a unique integer ID to every client, server, and page
appeared in the trace, and generates two output files which can be used to 
drive PagePool/ProxyTrace. Its usage is:

	tr-stat <MaxReqNum> [<Duration> <StartTime>]

	MaxReqNum: the maximum number of requests in a trace file
	Duration:  the length of the trace to be converted.
		   Its default value is the entire trace.
	StartTime: the starting time of the trace where to start conversion.
		   Its default value is the starting time of the first 
		   record in the trace.

Note that the paths and libraries in these makefiles are set ad hoc. For 
different installations, users have to change the paths and libraries.

The output of 'tr-stat' consists of two files: reqlog, and pglog. 'reqlog' 
contains the request stream extracted from the trace, its format is:

	{<time> <clientID> <serverID> <URL_ID>}
	i <Duration> <Number_of_unique_URLs>

The format of pglog is

	<serverID> <URL_ID> <PageSize> <AccessCount>

For an example as how to use these files with PagePool/ProxyTrace, please 
look at ~ns/tcl/ex/simple-webcache-trace.tcl.