Mary's Gardens wrote:
> I am able to get this desired output from the weblog stats program,
> but since weblog is not supported by my server, and analog is, I
> would prefer to obtain it, or an acceptable equivalent, on analog.
>
> If, as you state, and the docs and FAQ's seem to indicate, the output
> I desire cannot be processed by the existing analog code, my request
> then becomes one to the author to consider writing some additional
> code to accomplish this - thus enlarging the capabilities of analog.
This multi-variate reporting requires huge databases -- especially since you
are choosing to cross-compare the hosts (likely the largest data table) with
requests (likely the third largest). The memory requirements for this type of
analysis would be exorbitant and would degrade the performance of Analog
immensely and rapidly.
This is discussed in docs/faq.html item B18.
> While the requested output is not of much interest to commercial
> websites seeking to correlate sales with hits, it is very important
> to instructional/inspirational websites such as ours, which, for the
> continuing perfection of home page arrangement and file title
> selection, wish to examine qualitatively the e-trails by which hosts
> move, or don't move, sequentially from requested file to requested
> file.
These may also not correspond to "trails" of traffic by your users. Any given
host from AOL will be a set of lateral requests across many users (AOL's proxy
server request all parts of a page at once and AOL users go to the lowest
traffic proxy at each new request). So don't be fooled into thinking this may
have any bearing on traffic patterns or usage of your site.
This is discussed in docs/faq.html item B21 and docs/webworks.html.
> An acceptable equivalent analog output for these purposes - as the
> host files do not have to be alphanumerically sorted for this
> analysis, but could sequential - would be a sequential listing of
> hosts with added columns for
>
> date:time:request
>
> and everything else from the raw log non-selected
>
> In this, the equivalent for the sample above would be outputted as:
>
> 165.107.135.181 26/May/1999:18:38:35 /CMG-LC-1.gif
> 165.107.135.181 26/May/1999:18:38:35 /CMG-LC-2.gif
> 165.107.135.181 26/May/1999:18:38:35 /JS-PCMG-MG.html
> 165.107.135.181 26/May/1999:18:38:36 /CMG-LC-3.gif
>
> which would still give the desired sequential "e-trail" of each
> host's requests, albeit with repetition of the host URL.
>
> Due to the numbers of requests, such analysis would only be made by
> us on occasion and for a brief time interval - viz. of one day for
> our small, 500 host accesses a day, volume.
And the finally point (IMHO): Do you really look at the traffic patterns of 500
hosts every day? The point of Analog (and all other web log analyzers and
similar statistics systems) is to provide summary information to help you
understand large volumes of data. Sure it would be nice to be able to
'drill-down' to each host, but how often do you have time to do this and what
does it gain you? Do you Actually look at the entire host report or only the
top 50? For most of the report I configure, we don't even use the host report.
We're not really interested in what the specific host IPs were, just aggregate
summaries, like the domain and organization reports.
Jeremy Wadsack
Wadsack-Allen Digital Group
------------------------------------------------------------------------
This is the analog-help mailing list. To unsubscribe from this
mailing list, send mail to [EMAIL PROTECTED]
with "unsubscribe" in the main BODY OF THE MESSAGE.
List archived at http://www.mail-archive.com/[email protected]/
------------------------------------------------------------------------
