> My  mail servers are not there for Verizon to block their spam with,
> nor anyone else that wants to use SAV.

Your  argument  is  only  consistent  if  you believe that domain-only
validation   (rather   than   user/LHS   +   domain/RHS)   is   *also*
inappropriate,  since  aren't your _DNS_ servers *also* "not there for
Verizon to block their spam with"? For that matter, what gives you the
right  to  check  to  see  if  a connecting IP that forges the HELO of
mail.example.com   passes   a   HELO-A-PTR   roundtrip?  Doesn't  that
necessitate the unilateral use of example.com's DNS servers, which are
"not there for" your internal spam measures? And the list goes on.

> Yesterday we were connected to 898 times by Verizon to validate 
> addresses, and 697 of those connections were for addresses that don't 
> exist.  To make matters worse, when they get a 550 for a bad address 
> using a null sender, they RSET and try again using a named Mail From, so
> they actually tried bad addresses 1,394 times in just one day.  Who has
> a right to pound a server with 1,394 bad addresses a day?

Pound  with 1,394 RCPTs? You should be able to handle millions of RCPT
lookups  per  day,  or  even  per hour, without blinking! When you can
handle  the  traffic and are just trying to keep your addresses out of
spammers'   hands,  it  can  be  frustrating  to  try  to  use  simple
anti-dictionary  attack  logic  in  a  complex world; I can appreciate
that.  But  this  whole pity-me logic for a couple thousand lookups is
kind of weird.

> The  RFC  suggests  that  3 minutes [to respond to SMTP commands] is
> reasonable,  and  to  go  all  the  way down to 30 seconds is really
> pushing  it  in  a  world  where  there  is both tarpitting and many
> servers with performance issues due to spam.

If  you  can't  respond to RCPT TO validation in less than 30 seconds,
that  sucks  for  you,  but it's not representative. OTOH, if you just
*won't*  respond  to  RCPT  TO validation in that time, I doubt you're
really  benefiting  from  purposely  tying  up a socket for 3 minutes;
tarpitting  usually  ends  up  shooting  servers  in  the  foot unless
tarpitting,  like  everything  else,  adjusts  intelligently  to  true
threats.

Anyway,   the   less  forged-sender  spam  gets  accepted,  the  fewer
performance  demands  there  are  for  that  spam.  The  aggregate CPU
utilization of an SAV-based rejection, including the inbound HELO/MAIL
FROM/RCPT  TO  and the subsequent callout, is far less than what would
be used for content scanning. That's why SAV is done. Yes, the bulk of
the  resource savings is on the receiving side of any transaction; the
direct  achievement  of  SAV  is  protecting  the  receiver's  MX  and
mailboxes  from  abuse,  without  any inherent regard for the physical
resources  of  the  sender domain. But the indirect result is to force
visibility  --  either  through  blatantly  fraudulent  JJ  of  extant
[EMAIL PROTECTED], or through the use of spammer-only domains -- rather than
letting spammers hide behind unattended/nonexistent addresses.

Again,  I'm  not  saying  SAV is without problems, but you're throwing
some  FUD on the fire here... considering that just yesterday you were
characterizing Verizon's actions as something completely different and
totally without precedent.

--Sandy



------------------------------------
Sanford Whiteman, Chief Technologist
Broadleaf Systems, a division of
Cypress Integrated Systems, Inc.
e-mail: [EMAIL PROTECTED]

SpamAssassin plugs into Declude!
  http://www.imprimia.com/products/software/freeutils/SPAMC32/download/release/

Defuse Dictionary Attacks: Turn Exchange or IMail mailboxes into IMail Aliases!
  
http://www.imprimia.com/products/software/freeutils/exchange2aliases/download/release/
  
http://www.imprimia.com/products/software/freeutils/ldap2aliases/download/release/

To Unsubscribe: http://www.ipswitch.com/support/mailing-lists.html
List Archive: http://www.mail-archive.com/imail_forum%40list.ipswitch.com/
Knowledge Base/FAQ: http://www.ipswitch.com/support/IMail/

Reply via email to