On 2014-01-30 16:57, Travis Dolan wrote:
The single VM per client option is a good idea. However when dealing
with 100s or 1000s of clients management of those VMs becomes "fun".
We roll our MailPiler install out via AWS, and we manage the
configuration via Puppet. We can roll out as many machines as needed,
for as long as they are needed very effectively with little time on
behalf of our DevOps team.
Availability is achieved simply by snapshotting the disks used to
store the mail/db and then attach those disks to new EC2 instances in
the event of outages, development purposes, or fir additional
resource. Consistent snapshots are crucial here and are achievable via
many different tools available online.
This month alone we have sent over 8 million messages to one
MailPiler instance with 0 issues.
that's a success story I always wanted to hear :-)
Like anything in this business there are always many ways to achieve
the same goal. Our design works great for us, and maybe there are some
ideas in there to help others.
well, that's what a mailing list is for. Anyway thanks for sharing your
opinion.
Janos
