You have two options. Create a custom solution or use a commercial one. I ended up having to create a custom solution for the 85k mailboxes I manage.
Having investigated scalable mail systems I would recommend at least taking a look at Mirapoint. They aren't perfect but they are professional and have a very nice solution, though it will cost you.
If you're going to do things yourself I'd suggest looking at some of the following:
RedHat Enterprise Linux 4 Has most of the following already installed and you'd want to subscribe to a version of RHN that let you rapidly roll out new servers or upgrades/security patches to existing servers
OpenLDAP Can be used for authentication and directing mail and pop/imap or even webmail session to the appropriate backend mail stores.
Perdition POP/IMAP proxy can use LDAP
Postfix Again can use LDAP
Apache with PHP I used this to proxy/redirect webmail logins
Webmin It's cluster feature is actually quite handy and it's monitor scripts along with some Perl make for a quick and easy monitoring solution.
Using the above you can setup front end mail exchangers doing various anti-spam and anti-virus work in a load balanced setup with dynamic banning of IPs based on logs of refused mail. They should make use of LDAP so you don't allow any mail in that is destined for a non existant user.
Then you can use this to balance multiple back end servers of virtually any description. You could even have multiple vendor solutions used for the backend servers. Of course you'll need to tie it all together with custom administration scripts, etc
I WAS reporting scans and probes of my networks on a daily basis.
I'd semi automated this and at one stage was one of those that emailed Kirk without even realising that I had autorpm running on all my Linux boxen and that was what was triggering it.
I'd prefer to think of myself as over zealous rather then clueless.
I've since given up on the whole idea. Yes I managed to alert some people to the fact their machines had been hacked and they were very thankful for it. However that was not the norm and the time spent sending the emails (even once semi-automated) could not be justified by the results.
The norm was no response at all and often the worst offenders are the BIG ISPs in that department.
Eg. Telstra have a customer that regularly hits my network with broadcasts to a certain port which is presumably a misconfigured Innoculan (anti-virus) client. Do you think Telstra would bother to reply to me or pass on my message to their customer... Not likely!
Anyway in answer to Kirks question yes this is probably going overboard and admins should probably look at a combination of firewall logging and an IDS like snort to spot the true hostile activity.
I recently began running snort here and whilst I still don't bother reporting things at least I now have a better idea of what is thrown at my network each day and a MUCH better chance of picking up an attempted hack.
So far the most common malicious thing I see is an attempted exploit of LPRng for RedHat 7.0
I'll stop babbling now and Kirk you have my apologies for ever bothering you and my thanks for a great program.
Slashdot Mirror
Having investigated scalable mail systems I would recommend at least taking a look at Mirapoint. They aren't perfect but they are professional and have a very nice solution, though it will cost you.
If you're going to do things yourself I'd suggest looking at some of the following:
Has most of the following already installed and you'd want to subscribe to a version of RHN that let you rapidly roll out new servers or upgrades/security patches to existing servers
Can be used for authentication and directing mail and pop/imap or even webmail session to the appropriate backend mail stores.
POP/IMAP proxy can use LDAP
Again can use LDAP
I used this to proxy/redirect webmail logins
It's cluster feature is actually quite handy and it's monitor scripts along with some Perl make for a quick and easy monitoring solution.
Using the above you can setup front end mail exchangers doing various anti-spam and anti-virus work in a load balanced setup with dynamic banning of IPs based on logs of refused mail. They should make use of LDAP so you don't allow any mail in that is destined for a non existant user.
Then you can use this to balance multiple back end servers of virtually any description. You could even have multiple vendor solutions used for the backend servers. Of course you'll need to tie it all together with custom administration scripts, etc
Or "sudo vi" or "sudo less" or anything else that lets you invoke a shell.
I WAS reporting scans and probes of my networks on a daily basis.
I'd semi automated this and at one stage was one of those that emailed Kirk without even realising that I had autorpm running on all my Linux boxen and that was what was triggering it.
I'd prefer to think of myself as over zealous rather then clueless.
I've since given up on the whole idea. Yes I managed to alert some people to the fact their machines had been hacked and they were very thankful for it. However that was not the norm and the time spent sending the emails (even once semi-automated) could not be justified by the results.
The norm was no response at all and often the worst offenders are the BIG ISPs in that department.
Eg. Telstra have a customer that regularly hits my network with broadcasts to a certain port which is presumably a misconfigured Innoculan (anti-virus) client. Do you think Telstra would bother to reply to me or pass on my message to their customer... Not likely!
Anyway in answer to Kirks question yes this is probably going overboard and admins should probably look at a combination of firewall logging and an IDS like snort to spot the true hostile activity.
I recently began running snort here and whilst I still don't bother reporting things at least I now have a better idea of what is thrown at my network each day and a MUCH better chance of picking up an attempted hack.
So far the most common malicious thing I see is an attempted exploit of LPRng for RedHat 7.0
I'll stop babbling now and Kirk you have my apologies for ever bothering you and my thanks for a great program.
I'm a SysAdmin at a small Melbourne based ISP.
My best advice would be think again and if you still want to do it then think again...
Do a LOT of research and create a solid business plan. By the time you do that you'll probably see why I'm saying think again.
If you still want to do things and want some specifics on servers, etc feel free to email me (just remove the NOSPAM.)
Some of the info about EQ and hence the possibilities for the new game being presented here are somewhat innacurate so take them with a grain of salt.
:)
Eg. EQ is highly dependant on Direct X.
Considering I play it in Glide this is a bit misleading...
or Server loads suck and you have 2000 ppl playing on one server.
The EQ world is broken in to zones and each zone actually runs on a server in a server farm. What breaks it is too many ppl in one zone.
Anyway, just had to add my bit.
I'm not sure what it does yet. Have to get my box at home here up and running with Linux again.
l ive-0.1b.tar.gz
Check out:
http://developer.soundblaster.com/linux/
For this file:
ftp://ftp.soundblaster.com/pub/creative/beta/sb