[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Mail Server Design



My backup strategy, is every day at 3AM, y make a tar of
/var/spool/mail. and passes to another server. There is no need of
having mirrors, because users knows that mail server could crash, and i
will be restore on 4 hours.
I make Spam filtering / AV checks, on other SMTP server, and then it
passes to this server.

On Fri, 2004-04-16 at 10:12, Rory Campbell-Lange wrote:
> A couple of comments:
> 
> What is your backup strategy? On the physical hardware side you may wish
> to look into supporting a redundant disk over RAID for at least the
> single server approach, to help deal with the inevitable disk crash :).
> If you want to mirror numbers of other servers for backup purposes you
> may have problems with synchronisation unless you do it through backup
> rather than, say, mirroring via your smtp daemon, as imap/pop3 usage
> will obviously change the contents of each mail area.
> 
> You need to choose a mail daemon that can easily farm off subsets of
> users to seocndary servers, if you wish to go for version 2 scenario.
> Exim4 seems to me to be a good choice.
> 
> Rory
> 
> On 16/04/04, Mariano Wahlmann (wahlmann@agro.uba.ar) wrote:
> > I need some help, on designing a mail server solution (structure), my
> > request are:
> > 
> > -Users 4000+ aprox. with a growth rate of 20/users by month.
> > -Each user has 15Mb of inbox.
> > -Accesible by POP3 / IMAP (Only for Webmail who resides in other server)
> > -Mail list (some of 4000+ users)
> > 
> > I need a design solution for the server in wich resides mailboxes.
> > 
> > I have two proposal:
> > 
> > 1- One big expensive server with 1 SCSI disk, two processors.
> > 2- Several cheaps servers, dividing mailboxes across servers.
-- 
_________________________________________________
Mariano Agustín Wahlmann
Administrador de Red
Facultad de Agronomía - Buenos Aires - Argentina
Te.: (+54 11) 4524-8000 int.8108
email: wahlmann@agro.uba.ar
www: http://www.agro.uba.ar




Reply to: