Re: spam blocking experiences with MXs

["Joseph C. Bender" <benderjc_(_at_)_benderhome_(_dot_)_net>]

At 10:51 AM 12/21/2002 +0100, Robert Waldner wrote:
>On 21 Dec 2002 04:33:10 GMT, "Darren Reed" writes:
>Ok, let's do the math: the average piece of spam, based on what I have 
> in my archive, is about 6 kB. Take a few bytes out because of the 
> locally generated received:-lines et al, say, it's about 5.800 bytes
> now.
>
>Add to that IP-, TCP- and SMTP-overhead, say, about 1.500 bytes per 
> SMTP-transaction.
>
>So, delivering the spam instantly would cost you about 7.300 bytes, 
> retries every 30 minutes for 5 days amounts to 45 kB or roughly 6 
> times as many bytes.
>
	Your math is slightly off.

The MX-bouncer daemon doesn't even get that far.  It does a 450 after the
redirect happens in pf.  So you'd just be doing the smtp protocol overhead
up to that point.

I do think that Darren's got a point with the retries, but won't most MXen
be configured on a increasing backoff schedule?  
I thought Sendmail did that out of the box, so that clueless mail admins
wouldn't have their MXen pounding on other MXen?


Signing off, 

Joseph Bender
benderjc (at) benderhome.net
This account is used primarily for reading and responding to mailing list
traffic and is not my main mailing address.