What is required when deploying co-resident VMs by using Cisco UCM?

Cisco’s “Collaboration Virtualization – Application Co-residency” policy states that, when multiple Cisco UC applications (for example, Cisco UCM co-resident with IM&P, Unity Connection, etc.) are placed on the same physical ESXi host, hardware resources must be fully reserved for every VM. CPU, memory, and storage I/O summed across all VMs may never exceed the server’s physical capacity; i.e., no hardware oversubscription is permitted. This rule ensures deterministic real-time performance that voice workloads require. The other listed items (10 Mbps bandwidth guarantee, running VMs on an existing UCM server, or relying on application-level QoS) are not cited as prerequisites for co-resident VM deployments.

A. Bandwidth guarantees relate to network design, not the virtualization requirement for co-resident UC VMs.
B. VMs are deployed on a VMware ESXi hypervisor; the host itself does not run UCM natively.
D. QoS marking/queuing is handled by the network; it is not stipulated as a mandatory condition for co-resident VM support.

1. Cisco Systems, “Collaboration Virtualization – Application Co-residency Support Policy,” section 3.2 “Resource Reservation,” pp. 8-9.

2. Cisco Unified Communications Virtualization Guidelines (DOC-ID: UCS-UC-VIRT-GUIDE-12), section 5.1 “No Hardware Oversubscription,” p. 37.

3. Cisco Validated Design, “Unified Communications on UCS,” CVD-UC-UCS-2019, chapter 4 “VM Placement Rules,” p. 4-7.

(All documents available at https://www.cisco.com; access requires Cisco login.)