[TriLUG] seeking advice for a bullet-proof mail server

[Aaron S. Joyner]

Ryan Leathers wrote:

>My problem there Joseph is that I won't have the luxury of waiting.  In a
>nutshell, this mail server is a component of a business process automation
>plan.  Some number of the mail messages arriving will be the type that must
>be delivered in near real time.  That's why I mentioned using LVS to
>distribute incoming mail, or MX balancing. Its not about performance, just
>reliability.  If a mail server dies I have to keep chugging even while it is
>being recovered.  Queuing will not suffice except for the normal customer
>mail.
>  
>
There are some things that you simply have to wait on.  My other email 
is considered a pre-requisite to this, I won't attempt to explain 
again.  :)  If an MTA crashes, and a path to the back-end 
(client-accessed) server is still available, there is no perceived 
interruption of service.  If the back end server crashes, there is 
*going* to be a perceived interruption of service.  If you want to avoid 
that, by having redundant machines, you have a seriously up-hill battle 
ahead of you.  Put simply, it's better to put a bit more redundancy into 
that machine (RAID, good server-quality hardware, redundant power 
supplies) to ensure that it doesn't go down very often, i.e. close to 
never, and that's your best bet.

Aaron S. Joyner