Re: Ground.ecn download stats

[Rubywand <rubywand@swbell.net>]

"ground.ecn AppleII Librarian" writes ...
> 
> October 19, 2000
> 
> Jeff Fink asked me to do some download statistics on ground.ecn.uiowa.edu.
> This report summarizes _only_ FTP downloads (not http access).
> 
> I have run some summary stats for file downloads from ground 1996-2000.
> This turned into a challenge which I couldn't resist conquering.  I have
> created a Top100 list (actually top 107 since I'm using all files
> downloaded 300 or more times).  Some stats are a little misleading since a
> multipart file (ie some binscii files) is counted once for each part.  The
> Top100 file is available on ground in the main apple2 directory.  It is
> also posted as a separate message on csa2.
> 
> Challenges:  FTP log files        1996    6,254,776 bytes    83506 records
>                                   1997   35,007,752 bytes   442134 records
>                                   1998   56,929,482 bytes   688684 records
>                                   1999   39,239,234 bytes   456728 records
>                                   2000   76,906,213 bytes   859367 records
> 
> Processing time with SAS 6.12 on HP B132L workstation: 36 minutes
> 
> I eliminated 'duplicates' (file downloaded more than once on a given
> machineID) and a few files (name <4 chars (files in the lost+found)) or
> any of the 'FILE.LIST' files.
> 
> Top 107 list of file downloads from ground Feb 1996 - Sep 2000
> 
>                           Ground Usage - FTP downloads
>  Filename                    1996    1997    1998    1999    2000    Total
> 
>  index.html                    12     176     138      94     199      619
>  aboutdivdos.txt                .     190     139      85     145      559
>  ARCHIVE.SUFFIX.INFO.txt        .     170     177      88     116      551
>  spyhunter.1of3                17     208     141      91      94      551
>  ...
>  Lambda.txt                    21     102      68      43      67      301
>  ProTYPE11.hqx                  .      69      71      65      96      301
>  text.adventures               49      60      64      51      76      300
> 
>  Total downloads            72830  381949  478553  406191  740070  2079593
> 
>  Total filenames = 32978
> 
> As you can see from the Top100 file, over 2 million file downloads have
> occurred in the past 5 years.  In the first 9 months of this year over 740
> thousand downloads have occurred.  We may reach 1 million downloads for
> the year 2000 by the end of December.  (That is on average over 2700
> files/day!)

     That's Awesome!


> 
> A complete list of the 32978 files in the summary is 600 pages, single
> spaced. (Only 2,785,445 bytes.)
> 
> Determining the number of users who have downloaded files from the ground
> archives is nearly impossible.  The FTP logs contain 'machine id' instead
> of userid or login names.  There are over 27000 unique machine IDs.  But
> many of these represent (bogus) generated IDs. There are thousands of IDs
> that begin with [cisco,cvx,dial,du,host,ip,kmm,pm,port,ppp,proxy,oct,
> remote,rock,slip,sf,tc,tp,ts,user] to name a few.  AOL represents another
> situation because most aol users are identified as xxxxxxxx.ipt.aol.com
> (where xxxxxxxx is a RANDOM alphanumeric string not attached to a
> particular user).
> 
> So instead of users, I tried to identify 'sites' by stripping off the
> leading level of each machid (ie all chars up to and including the first
> '.', except when the machid was a IP number (www.xxx.yyy.zzz).  Thus all
> AOL users were reduced to one site.  Same for ppp logins from particular
> service provider, etc.  This reduced the unique 'sites' to 13560. Still,
> 4265 'sites' had downloaded only one file.  So only 9295 sites have
> downloaded more than one file.  4382 sites have only downloaded between 2
> and 5 files, thus 4913 sites have downloaded 6 of more files.  3584 sites
> have downloaded 10 of more files.  61 sites have downloaded over 10,000
> files apiece.  25 sites have downloaded over 20,000 files.
> 
> Given that some 'sites' represent more than 1 user, and assuming only
> those 'sites' with 6 or more downloads represent an active apple2 user,
> I estimate the number of apple2 users who have actively used
> ground.ecn.uiowa.edu to be over 5000 from all over the world.
> 
> A complete list of the 13560 sites in the summary is 247 pages, single
> spaced. (Only 1,145,520 bytes.)  The top100 list would include sites which
> have downloaded 5340 or more files.  I would consider most of sites as
> private mirrors of (parts of) ground.ecn
> 
> Site              Rank    1996    1997    1998    1999    2000   Total
> 
> ex-pressnet.com      1       .       .       .   21513   13878   35391
> dialnet.net          2      23   30416    4373       .       .   34812
> omah.uswest.net      3       .       .       .     193   33411   33604
> ...
> idx.com.au          99       .    5411       .       1       .    5412
> deltasoft.com      100       .       .    5340       .       .    5340
> ...
>                           ====    ====    ====    ====    ====    ====
>   N = 13560              72830  381949  478553  406191  740088 2079611
> 
> I believe these statistics demonstrate that ground.ecn.uiowa.edu continues
> to be one of the premier apple2 sites for a world-wide group of users.
>

     For sure. They show that Ground is a lot busier than I ever imagined.


 
> Thank you to Jeff Fink (author of Silvern Castle adventure game) for
> nudging me into this project.  And a special thank you to apple2 users
> world-side for supporting ground.ecn.uiowa.edu
> 
>   --Steve


     Thanks for keeping Ground on-line and growing!


Rubywand