Web Statistics

[Bill Feidt]

An interesting observation about Web server statistics:
Our statistical breakdowns reveal that more than 25% of our
hits come from "unresolved" IP addresses.  These are IPs 
that are not registered in the DNS system and therefore 
cannot be resolved to names.

I suspect there are many interesting facts behind those 
unresolved addresses.  For example, this morning I noticed a
number of hits from the IP: 166.111.72.35.  Investigation
showed that this IP belongs to Tsinghua University in Beijing.
Further investigation showed that this IP had accessed our Web 
server 2,374 times during April.

On the other hand, ".cn", the top level domain name for the
PRC, accounts for only 165 April hits.  Add the unresolved
Tsinghua accesses, to the ".cn" hits and PRC activity accounts 
for a far greater percentage of resolved accesses than any of: 
Germany, Italy, Netherlands or Japan. In fact, it trails only 
the U.S., Canada, Australia, and the UK in terms of access.

Further inspection of our logs revealed that most of the 
166.111.72.35 accesses had the hallmarks of a robot indexing 
run, in my mind, somewhat lessening their significance.

Of course, the foregoing is a misleading anyway since we don't 
know how many unresolved IPs correspond to accesses from Germany, 
Italy, Netherlands, Japan, etc.

It's just something to keep in mind when considering Web
statistics.  At a surface level, they often conceal as much as 
they reveal.

Bill Feidt
wfeidt at nal.usda.gov
AgNIC
<http://www.agnic.org>