This is what the Analog cache files are for (see
http://analog.cx/docs/cache.html). You can create a cache file for
each day's results and then write reports by reading in the cache
files. This will certainly improve performance. It may or may not help
memory problems.

If you are running out of memory, you can try using the *LOWMEM
commands (see http://analog.cx/docs/lowmem.html). Generally the Host
and Referrer data are the most memory consumptive. You can also add
more memory to the box running Analog (memory is pretty cheap these
days).

You can't just run reports and use those to produce other reports
because they don't contain complete information. Because the reports
only show the top items, they will not properly count items not
included in particular reports.

Once you have exhausted all these options there are a couple
additional things you can try. We configured one large client's
processing with very careful use of *LOWMEM 3 commands (which remove
data from the reports). You can run a successive series of reports on
each data type and then recombine them. This will take a long time
(about 5 - 6 times as long as the original processing) but uses less
memory. It's imperative, however, that you do conversion testing on
known data to guarantee that you are gathering and calculating
everything correctly in the process.

A more effective approach is to look at your data and make some
decisions about what you *really* need to report on and what you
don't. For example, how useful is and annual report? Do you need to
track all the graphics files in your reports (bandwidth tracking is
better handled through network monitoring apps anyway)? Can you get
more effective reports by splitting the data into domains based on
target audiences or user groups? Finding effective ways to implement
*INCLUDE/*EXCLUDE and *ALIAS command to reduce the data volume that
Analog processes can produce long-term (assuming a growth trend)
approaches to data management.



-- 

Jeremy Wadsack
Wadsack-Allen Digital Group


Monsur Hossain <[EMAIL PROTECTED]> (Tuesday, September 09, 2003 8:27 AM):


> Not sure about doing this through Analog, but I wrote a helper app to kick
> off analog.  Once analog is done running, the helper app uses regex to parse
> the output html and grab the relevant data I wish to track over longer
> periods of time (successful reqs, redirected reqs, failed reqs and data
> transferred).

> Monsur



> -----Original Message-----
> From: [EMAIL PROTECTED]
> [mailto:[EMAIL PROTECTED] On Behalf Of Cal Lidderdale
> Sent: Monday, September 08, 2003 4:59 PM
> To: [EMAIL PROTECTED]
> Subject: [analog-help] File size problem


> Sounds too good to be true - its not what one would think - we're using
> Apache as a data server  [ ie: vox files et-al ] and the requests run in the
> 1/2 Mil / day range with todays access_log at 60 Mb and yesterdays at
> 60 Mb and ....   Like da guy sayz  "Billion here, billion there, pretty
> soon we'er talk'n real money."

> So my question is:  Is there a way to run "a" day  summary that would be
> equal to a one day web page and store it?  FF or DB.  Then when I run
"analog .... >> htdocs/Stats/THE-WHOLE-YEAR.html" it would combine each days
> "summary" into a report.  This way it a) wouldn't take 3 hrs and b) won't
> say "out of memory"  :'(

> Tks
> Chuck

+------------------------------------------------------------------------
|  TO UNSUBSCRIBE from this list:
|    http://lists.isite.net/listgate/analog-help/unsubscribe.html
|
|  Digest version: http://lists.isite.net/listgate/analog-help-digest/
|  Usenet version: news://news.gmane.org/gmane.comp.web.analog.general
|  List archives:  http://www.analog.cx/docs/mailing.html#listarchives
+------------------------------------------------------------------------

Reply via email to