Interesting new spam technique - getting a lot more popular.

chuck goolsbee chucklist at forest.net
Thu Jun 15 20:36:16 UTC 2006


At 2:35 PM -0400 6/15/06, Matt Buford wrote:
>But how could this possibly be IP abuse or evil (except perhaps in 
>the eyes of the search engines)?  What difference does it make to 
>ARIN if I give a customer 30 IPs from a single /24 or 30 IPs from 30 
>different /24s?

How is that customer using those IPs? If the IPs are on a single 
server used for webhosting, it is in violation of ARIN's IPv4 
allocation policy.

In every case where we've seen people asking for outrageous amounts 
of IP space for webhosting it is either because:

* They are trying to game the search engines due to this pervasive folklore.
or
* They lacked sufficient clue to grok name-based virtual hosting.

The latter can be fixed quite easily. I wish I had some way of 
debunking the former.

>It makes little difference to me and is trivial to do in my topology 
>since I already have 30+ /24s on the interface.

Just becasue you can, doesn't mean that you should. But hey, your 
network, your rules I guess.


>It is slightly more work to document the IPs since they each have to 
>be put into my database instead of a single range, but this is 
>handled by the server people.

I prefer to have our 'server people' and our 'network people' working 
together without annoying each other too much.



While my use of the word "evil" was a smirking poke at the dominant 
search engine, I don't really think this behavior is malice so much 
as disregard for the ecosystem. We've done our best to be very 
conservative in our IP allocations to our customers, if nothing else 
to remain good neighbors to the rest of the Network.

I wasn't even aware of this bizarre SEO/IP scheme until we made that 
acquisition two years ago. Now I look around and see operations a 
fraction of our size consuming large allocations for small 
installations. The pursuit of a page rank seems a pretty selfish 
reason to consume a limited resource.



--chuck



More information about the NANOG mailing list