Orin,

> This  week  the  server farm where we have our iMail server suffered
> some  significant  problems...In  the  muck up we had our server out
> intermittently...

Luis has outlined the most common kind of redundancy, which is offered
by the DNS itself in the form of multiple MX records; he mentions that
having  two MXs at the same colo wouldn't solve colo-wide issues (such
as those you had). Rick is strong on the WAN side and has given you an
idea  of  the  complications  of  setting up multiple colos--if you're
going  to  do  it  right, you have to know who's peering with whom and
where, who's on the brink of bankruptcy, etc.

Now  would  probably  be  a  good  time  to put up a store-and-forward
secondary MX, just to get used to the concept. Once it has a record in
DNS, you can move it to another colo as necessary just by changing the
published IP.

For  this  secondary, you could use MS SMTP or get another license for
IMail. Your secondary MX normally just holds on to messages until your
primary  comes  back up (usually with a short queue retry period and a
longer  queue  lifetime  to  reduce  the number of bounces back to the
originator), so no messages are ever delivered locally to that box. MS
SMTP, or any MTA, will thus suffice.

If  you  use  IMail,  you  have  the additional option of, in a pinch,
turning  your  secondary  into a mailbox server and letting your users
access their mail via the web. You'd do this by keeping a dummy domain
(say,   backup.example.com)   on  the  backup  server  and  frequently
synchronizing   it   with   your  main  mailbox  server  via  Registry
backup-and-restore.   In   a  longer-term  emergency,  you  could  add
example.com  as a host alias under backup.example.com, causing mail to
be  temporarily  delivered  locally  to the box. Of course, this would
only  apply  to  messages sent after the outage began. (A more extreme
version  of  this would be to set up both boxes as mailbox servers and
set  every  client  MUA  to  pull mail from both. This can cause major
setup  and  management  headaches  and  is  not advised; one immediate
casualty  is  webmail, which can go out of sync even if there's just a
blip of downtime on the primary.)

Additional  redundancy  options  involve  real-time  mirroring of your
userbase  across  geographical  locations--things your budget probably
doesn't allow. Why don't you wait for a few days to see if things cool
off?  There's  often  a  panic  period in which The Boss wants this to
"never happen again"--then reality sets back in. :)

-Sandy


To Unsubscribe: http://www.ipswitch.com/support/mailing-lists.html
List Archive: http://www.mail-archive.com/imail_forum%40list.ipswitch.com/
Knowledge Base/FAQ: http://www.ipswitch.com/support/IMail/

Reply via email to