[170344] in North American Network Operators' Group

home help back first fref pref prev next nref lref last post

Re: misunderstanding scale, SMTP edition

daemon@ATHENA.MIT.EDU (Lamar Owen)
Wed Mar 26 13:45:46 2014

Date: Wed, 26 Mar 2014 13:36:03 -0400
From: Lamar Owen <lowen@pari.edu>
To: nanog@nanog.org
In-Reply-To: <20140326170906.13315.qmail@joyce.lan>
Errors-To: nanog-bounces+nanog.discuss=bloom-picayune.mit.edu@nanog.org

On 03/26/2014 01:09 PM, John Levine wrote:
> Quite right. If I were a spammer or an ESP who wanted to listwash, I 
> could easily use a different IP addres for every single message I 
> sent. R's, John 
Week before last I saw this in great detail, with nearly 100,000 
messages sent to our users per day from probably the same spammer (lots 
of similarities, including an image payload with invisible anti-bayesian 
text and a .in TLD) where no two messages came from the same IP.  It did 
all come from the same hosting provider, though, and at least for now 
that hoster's whole address space (all twenty blocks, varying between a 
/23 and a /17) is in my border router's deny acl for incoming on port 
25.  At least for now; I did send an e-mail out to the abuse contact, 
waited 72 hours, then but the blocks in the incoming acl.  This hoster 
was adding rwhois entries for each /32 allocated (yes, IPv4 /32) and 
they had different NIC handles.  I'll probably wait a month, then pull 
the acl to see if it starts back up.  Oh, and each and every /32 that 
sent mail had fully proper DNS, including PTR etc.  Spamassassin's score 
was well in the 'ham' category for all of those messages.

IP reputation lists are one weapon in the arsenal, but not nearly as 
effective as one would like.  There is no technical magic bullet that 
I've seen work over the long haul.

But that's not really on-topic for NANOG.



home help back first fref pref prev next nref lref last post