Two weeks. For a hotel chain with 24/7 operations across eight time zones. That wasn't an option.
She launched the Cisco-provided OVA template for CUCM 12.5. Four vCPUs. 8GB RAM. 110GB thick-provisioned eager-zeroed disk. The UCS blades hummed as the VM materialized on shared storage. No local disk failures. No proprietary hardware dependencies. Just pure, clean compute. cucm virtualization
The problem? Their legacy Cisco Unified Communications Manager (CUCM) cluster—three physical MCS servers, affectionately nicknamed "Big Yellow," "Old Blue," and "The Grouch"—had finally given up. Big Yellow had suffered a catastrophic RAID failure at 4:00 PM. The vendor quoted two weeks for a replacement part. Two weeks
CUCM's virtualized heartbeat timers are notoriously sensitive. In a physical world, a 200ms delay is a shrug. In a hypervisor, if the ESXi host gets busy, that same delay can trigger a "node isolation" event. The cluster would split-brain faster than you could say "call manager group." She launched the Cisco-provided OVA template for CUCM 12
It was working.
The future of voice wasn't in beige boxes anymore. It was in a few gigabytes of RAM, a reservation policy, and an engineer who knew when to break the rules.