Production servers must stay online. Hardware and networking failures are among the primary causes of downtime.
Wouldn't it be nice to have a system that reliably failed over to another server when Moore's law strikes? Wouldn't it be nice if the standby server takes over without interruption of service to the users?
Since the Linux kernel 2.6.33, a system for replicating the server's storage comes built in. The Distributed, Replicated Block Device, or DRBD, replicates a system's mass-storage "block for block." DRBD, along with the OSS Heartbeat and Pacemaker, lets your server stack automatically failover to a standby should the primary server fail.
You may find a draft of the presentation slides here. You will note that my visuals are highly graphical--the feedback I receive from attendees tells me that audiences prefer a fluid and well-prepared presentation rather than just reading off slides.
I look forward to presenting at LinuxFest NW. If you would like me to present at this year's Fest I would be so happy to do so--I look forward each year to coming up to Bellingham!
Very Truly Yours,
D. Cooper Stevenson