UCS Network Configuration for UCCE
Design Considerations for High Availability
The High Availability (HA) deployment provides redundancy from a component failure in the Unified Computing System (UCS) architecture.
For deploying Cisco Unified Contact Center Enterprise (UCCE) on UCS B-Series servers, it is mandatory that HA must be used. Please follow the best practices and guidelines described in the Cisco white paper Design Consideration for High Availability and Scalability in Blade Server Environment for guidelines on implementing HA. Pay particular attention to the following information in this white paper:
- The guidelines described in the section “Server I/O High Availability Architecture” with redundant infrastructure and paths.
- The diagrams of HA design shown in Figure 2, Figure 4, and Figure 6 in the above white paper. For an additional example of Cisco UCS 6100 Series Fabric Interconnects HA, refer to Figure 5 in this document.
This HA design requires redundant Cisco UCS 6100 Series Fabric Interconnects on a Cisco UCS and redundant Cisco UCS 2104XP Fabric Extenders.
For an introduction to UCS architecture, see the Cisco white paper http://www.cisco.com/en/US/prod/collateral/ps10265/ps10281/solution_overview_c22-522771.pdf.
Setting up the Cisco UCS Fabric and the Ethernet Network
Figure 1 below shows the cable connections between the following UCS Fabric components with redundant configuration for a UCCE on UCS B-series deployment:
- Two Cisco MDS multilayer fabric switches
- Two Cisco UCS 6100 Series Fabric Interconnects
- Two Cisco UCS 2104 Series Fabric Extenders
- Dual redundant Ethernet Uplinks that will be connected to redundant Network Routers.
Please note that the B200 M1 blade is configured with the Cisco UCS M71KR-Q QLogic network adapter that provides 10 Gigabit Ethernet and Fibre Channel connectivity for the server. However, when deploying Clustering Over the WAN with B-Series hardware, the Cisco UCS M81KR Virtual Interface Card is used instead of the Cisco UCS M71KR-Q.
Figure 1 illustrates UCS B-Series network configuration for UCCE deployment.
UCS Fabric Ethernet switching mode must be set to End-Host Mode. Switching Mode is not supported.
Redundant chassis is not required, but recommended.
Figure 2 illustrates the UCCE Public and Private network connections set up in the UCS B-Series system for a multi-fabric deployment.
Figure 3 illustrates UCCE Public and Private network connections set up in the UCS B-Series system with split fabric (i.e., a deployment containing redundant interfaces for each network alternating Fabric side (A/B)) and Clustering Over the WAN deployments.
Creating the UCS Profile
Understanding the service profile is key to understanding the blade management in the UCS. The service profile represents a logical view of a single blade server, without needing to know exactly which blade you are talking about. The profile object contains the server personality (identity and network information). The profile can then be associated with a single blade at a time. The following link provides a step by step guide on how to provision service profiles.
Every server that is provisioned in the Cisco UCS is specified by a service profile. A service profile is a software definition of a server and its LAN and SAN network connectivity. In other words, a service profile defines a single server and its storage and networking characteristics. The UCS Manager software is embedded in the Cisco UCS 6100 Series Fabric Interconnects, which is also where the service profiles are stored. When a service profile is deployed to a server, UCS Manager automatically configures the server, adapters, fabric extenders, and fabric interconnects to match the configuration specified in the service profile. Automating the device configuration reduces the number of manual steps required to configure servers, network interface cards (NICs), host bus adapters (HBAs), and LAN and SAN switches.
This page does not provide a step by step guide on how to use UCS manager. However, it provides the necessary information to configure a service profile for UCCE.
VLAN Configuration for UCCE Public and Private Traffic
The service profile must have the complete and correct VLANs configured for UCCE Public and Private traffic.
The following shows an example of the VLANs needed for a UCCE deployment.
- One VLAN is needed for the Public Ethernet traffic
- One VLAN is needed for the Private Ethernet traffic.
Public and Private Traffic for Multi-Fabric Deployment
The following figure illustrates the VLANs needed on the SideA blade for most UCCE deployments. In these deployments, a similar VLAN configuration would be needed on the SideB blade.
Public and Private Traffic for Split Fabric and Clustering Over the WAN Deployment
For split fabric and Clustering Over the WAN deployments, the Public and Private networks must be configured to use their own redundant interfaces, with redundant interfaces for each network alternating Fabric side (A/B). It is also best practice to configure separate redundant interfaces for the management networks. These interfaces will be used in configuration of the vSwitches for networks in the HyperVisor.
Note MAC addresses in UCS Manager for all vNICs, as the vNIC numbers do not map to same number vmnic in the HyperVisor.
Note that in order to provision more than two interfaces per blade, the M81KR “Palo” Cisco Converged Network Adapter (CNA) must be used. The M71KR is only capable of provisioning two interfaces and is limited to the configuration shown in Figure 4.
The figure below illustrates redundant interfaces for Public, Private and management networks (where two pairs of Public interfaces are shown due to the example configuration having the management network on the same VLAN).
The service profile must have a vSAN configured for UCCE. The following shows an example of the vSAN needed for a UCCE deployment. The vSAN network ID must match the vSAN ID configured on the SAN devices for connectivity.
Configuring the VLAN on the Network Router
The following is an example of the port configuration for the network router. For redundancy there must be at least two TenGigabitEthernet ports configured per UCS 6100 Series Fabric Interconnect.
interface TenGigabitEthernet2/1 description Fabric-Connect-A switchport switchport trunk encapsulation dot1q switchport trunk allowed vlan 1,10,11 switchport mode trunk no ip address mls qos trust cos spanning-tree portfast trunk
Where VLAN trunking is used to uplink to VMware, it is essential that the network configuration be implemented correctly. Testing has shown that improper configuration of VLAN trunking directly and negatively impacts performance of the system.
As with all previous UCCE deployments, there must be dedicated public and private links between routers following supported models as defined in the Cisco Unified Contact Center Enterprise Solution Reference Network Design (SRND).
Configuring the Virtual Switch and vmnic Adapters
For deploying Cisco UCCE on UCS B-Series servers, configure the vSwitches and vmnic adapters as follows:
vSwitch and vmnic Adapter Configuration for Multi-Fabric Deployment
In most deployments, you create one ESXi virtual switch and connect both VM virtual Ethernet adapters to the virtual switch, as shown in the following figure. See Configuring the ESXi Host Network for the Virtual Machines for instructions.
NOTE: The Nexus 1000 virtual switch is not supported as a VMware switch.
vSwitch and vmnic Adapter Configuration for Split Fabric and Clustering Over the WAN Deployment
For split fabric and UCS Clustering Over the WAN deployments that use the M81KR Virtual Interface Card, create ESXi virtual switches and connect VM virtual Ethernet adapters as shown in the following figure.
Assign physical adapters (vmnics) to vSwitches by MAC address to ensure that the correct interfaces created by the UCS profile are being assigned. Make sure that the Public and Private networks use alternate Fabrics (A/B) on their Active vmnic.