MTA layer integration.. It scans at the end of the SMTP DATA phase prior
to accepting delivery. Messages can still be 4xx'ed or 5xxed at the end of the DATA phase, so this does in fact work.
Completely defeats the point of greylisting..
How so?
Last month (our) SpamAssassin (serverfarm) caught 103 million spam messages.
If you can block 70% of that with greylisting, does it really make sense to waste cpu(+ram) cycles processing / scanning and wasting bandwidth on garbage?
Also, pre-queue scanning simply will not scale for large isps. Even a half-baked spam flood will cripple your entire MX layer.
Whitelisting is the key for greylisting. Whitelist your outgoing smtp servers, and other big isps in your country.. Looking through your logs for previous months to see which hosts and/or domains are sending you(r company) losts of mails will save you huge headaches..
Cami
