Usually, you get exactly what you pay for, so when you
skimp on device redundancy, that's exactly what happens.
Server virtualization has brought us efficient and cost
effective disaster recovery options.
Snapshots, v-motion, and storage replication are the underpinnings of
this redundancy. However, its incorrect
to apply these server techniques to virtualized
Application Delivery Controllers (ADCs). The
constraints of memory utilization, product licensing, and
time/complexity-to-recovery are nuances which uniquely apply to ADC devices and
keeps them from enjoying the same methods of redundancy that virtualized server
workloads leverage.
A redundant and available architecture always starts with TWO appliances. It's just not appropriate
to architect an enterprise quality high availability solution with less. It might look good on paper, but the recovery
time will clearly exceed estimates- and likely won't work at all.
In some business continuity architectures, 3 devices may be appropriate. And in the future, Spotted Clustering, across multiple data centers may change best practices, but that's still a ways away and will require additional licensing. In the mean time, the best practice solution should always start with a pair of devices in each data center.
