Re: follow-up on web server counts

["T. K. Cassidy" <tkc@KUENTOS.GUAM.NET>]

>Date: Thu, 24 Apr 1997 14:46:26 -0400 (EDT)
>X-Authentication-Warning: Tesla.Public.Lib.GA.US: [168.29.252.63] didn't
>use HELO protocol
>X-Sender: dtope@mailbox.gpls.public.lib.ga.us
>Mime-Version: 1.0
>To: plrsnet@csn.net
>From: Diana Tope <dtope@mail.gpls.public.lib.ga.us>
>Subject: Re: follow-up on web server counts
>Sender: owner-plrsnet@lists.csn.net
>Precedence: bulk
>
>TO ANY of you who have any specific information about what the state library
>or any of the libraries in your states may be counting re. use of
>technology....Please send me information.  The Steering Committee will be
>meeting June 9-11 and we will be working on trying to identify and define
>some elements for FSCS gathering.  Talk to your large and small libraries
>and see what they are doing in this respect.  We really appreciate any
>information you can share with us, because we want to be sure that what we
>end up with is something that most, if not all, of the libraries will be
>able to count.  THANKS!  Diana TOpe
>
>
>************
>At 01:17 PM 4/24/97 -0500, you wrote:
>>I think we may have a pretty good thread here, as I've already heard from
>>several of you.  Here is our webmaster's (Liz Forney) explanation of the
>>summaries used to count hits on our SILO server (our web server).
>>
>>My conclusion is that actual usage can be determined by subtracting the
>>number of hits on graphical files (gif, jpeg, etc.) from the total number of
>>files served up in response to HTTP (hypertext transfer protocols) requests.
>>
>>Liz does, I think, an excellent job of describing how the web works and the
>>way software can be used to track usage...
>>
>>Gerry Rowland
>>Iowa
>>
>>A "hit" is a successful response to an HTTP request to the Web server.
>>What this means, that every Web page requested from our Web server will
>>generate at least one hit.  Graphics images on the Web page generate
>>additional hits for a page.
>>
>>Usage counts are difficult to generate in a Web environment, as the HTTP
>>protocol is considered stateless and sessionless.  One person might visit
>>our web site and request several web pages, and generate several hits in
>>the process, even though they are one user.  In the SILO ILL system, I
>>added a session tracking module, that starts when a librarian logs in.
>>This way I can tell how many times a librarian logs in during a given  time,
>>and get some sense of the activity for that session (loans  initiated, loans
>>responded to).  This I feel is a little bit more  accurate way to detail the
>>usage of our site.
>>
>>For the non-ILL Web requests, every request is logged to the Web server
>>access log, and a success/non-success is noted in that log.  Every night  I
>>go through the Web server access log with a public domain program from  the
>>University of California called "wwwstat" and compile the number of
>>successful hits and byte counts of Web documents requested, as well as
>>track who is accessing our site by IP/DNS name.  At the end of every  month,
>>I take the entire statistics for that month, and archive it into a  file.
>>The current reporting period and previous month's statistics can  be found
>>on the SILO page from the SILO Administration link, and from  their the Web
>>Server Statistics link.
>>
>>Through April 23, I show that 306,052 files were delivered successfully
>>from this web site.  We are already over the average of 150,000 for "hits"
>>for  this month.
>>
>>There has been considerable discussion on the Web on how to more  accurately
>>track usage.  "wwwstat" is one of the leading public domain  programs that
>>track web site usage, and it was definitely one of the  early ones.  One
>>thing you have to be concerned about is the inflation of  the statistics -
>>and what they really say.  We don't make excessive use  of graphics on this
>>site, but most pages do include at least one graphic  image (almost all of
>>the ILL pages only have one graphic on them).  To  take for an example, if
>>someone requests an ILL page from the Web server,  two "hits" are generated,
>>one for the CGI/HTML page and one for the  graphic.  The primary pages have
>>more graphics, so a hit is generated for  the HTML page and one hit for
>>every graphic on that page.  So the State  Library home page would generate
>>1 hit for the HTML page and 14 hits for  the graphic images on that page, in
>>its current configuration.  As you  can see, "hits" might inflate the usage
>>just a bit.  The "hits" for the  GIF images are noted in the Total Transfers
>>by Archive Section on Web  server statistics page.  A more accurate
>>reflection of Web server pages  delivered might be had by adding all of the
>>files with the html/cgi  extension, and not include the gif files.  I
>>personally feel that I need  to know of all file types delivered, because I
>>as a system administrator  need to know the system time used to deliver all
>>files requested whether they be GIF/HTML/CGI, etc.
>>
>>
>Diana Ray Tope
>dtope@mail.gpls.public.lib.ga.us
>156 Trinity Avenue SW
>Atlanta, GA 30303-3692
>(404)657-6245
>

Slainte! Guam is 15 hours ahead of Eastern Standard Time (USA) --
Living' in the future -- ain't Life grand!

T. K. Cassidy - Children's Librarian /// Guam Public Library
                             Resident Writer, Storyteller &  Author!! :-)
                             tkc@kuentos.guam.net
                             http://www.guam.net/home/tkc/

1270 North Marine Drive, Suite 101-424
Tamuning GUAM 96911 U S A

 April - the IRS chose well whenthey chose  the month we start by honoring
fools and end by cryng "May Day!