On 22/08/2011 06:06, Murray S. Kucherawy wrote:
[mailto:owner-ietf-smtp(_at_)mail(_dot_)imc(_dot_)org] On Behalf Of Hector
Sent: Sunday, August 21, 2011 6:59 PM
Subject: Re: Mail Data termination
Is there a loading limit in your arsenal against DoS attacks? Is
there such a thing as resource limits even for a Modern OS? And is it
reasonable for the CS client to pass the cost burden on the receiver
to solve what is only a client problem?
I think you're ignoring the points others have made that a connection cache
benefits both sides, since establishing a connection is a (relatively)
expensive action at both ends, especially if SMTP AUTH or STARTTLS are in use.
It doesn't strictly benefit the sender.
It doesn't benefit the receiver anywhere near as much as the sender.
The sender can open 50,000 connections to 50,000 different mail servers,
and just hold 49,999 of them idle and send one message at once = very
low CPU/bandwidth load. The receiver can't, as easily, hold open 50,000
connections from 50,000 different clients and be able to handle up to
50,000 incoming messages simultaneously in a timely manner.
The sender knows which connections will be used at any time. The
receiver has to assume they may all be used at any one time, so has to
have the capacity to handle them all. So, it's a very unequal relationship.