Thats  a lot of questions to answer.... 

 

So I will try to answer from our perspective...

a.) Our "Failover" machine is physically in another location. 

     If the "main" box, can't be contacted within 5 minutes, the Failover machine kicks off.

     (how we move data between the two, is a different discussion.)

    ** We did check with SM, that as long as we have an Enterprise license AND one 

        machine will not be running, when the other is up, we are compliant with license.

 

b.) Told you # a.) first, because to support our process, we have two 

     "gate MX servers" that handle incoming mail.  We are, or will be, once we get a

      bug ironed out with SM, be using SM exclusively for this purpose.
 

c.) We have own process to dynamically control DNS to coincide with #a.) switches....

 

Our position is this:  Our email HAS to be available... not having a redundant 

process is not an option for us.   

~~~~~

In reflection... we actually practice BCP (business continuation procedures) .. 3 times a year.

I know we are anal... but we learn something new everytime.  Being in the mid-west our concern is more suffering from a natural disaster, where a specific site could be down for many hours, if not days.

~~~~~

To me, it really depends on your business requirements.  The cost for hardware and software is not monumental .. in comparison to effects of being unavailable ( in our estimation).

 

 

 

 

 

HA for SM is unfortunately not the easiest thing out of the box. We have a backup MX server (another SM install) that is the secondary MX for all of our hosted domains. It is in a different data center that our primary MX server. This way, anything can happen to the primary MX server (hardware failure, IaaS provider outage, network issue, etc, etc) and our own MX server will catch all of the incoming email. Having the backup MX gets the email into our infrastructure as quickly as possible and you will not be waiting on potentially non-compliant servers to resend your customers mail to them as some predetermined window that you do not know.

 

I would like to see better support for HA setups for SM and have multiple installs accessing the same domain folder structure. Like Bruce mentioned, his client has to FTP its files to the other DC and have a DNS script that moves records. In a perfect world, you'd have two mailbox servers public facing (or more) all access shared storage for emails. You can easily load balance servers with DNS or a load balancer but SM can't handle two servers hitting the same data store. You also have a ton of different options for file replication (aside from FTP) with Amazon, Azure, or nearly any of the SAN providers like EMC. This would be separate from your (hopefully) redundant edge gateways that are doing your edge security/spam.