Since the early to mid-2000s, I have been consistent in how I designed a survivable branch office. My best-practices solution consisted of one or more gateways, a call processor, and some local trunks. During normal operations, the branch office relied on the enterprise’s main processor for all call processing activity and the remote processor would sit idle until the primary system died or the WAN connection went down. The gateways would then attach themselves to the local processor and calls would resume.
My latest article for No Jitter explores my evolving thoughts on the survivable branch office. What do you think?