The physical network of a VSA cluster consists of Ethernet switches and network interface cards (NICs) that are installed on each host.
Physical Network Architecture
All networking in a VSA environment must operate at speeds of 1G or higher in order for the configuration to be supported.
All hosts in the VSA cluster must have two dual-port or four single-port network interface cards. You can use a single Ethernet switch for the VSA cluster network. To ensure network redundancy, you should use two Ethernet switches.
The following illustrations depict network redundancy in a VSA cluster with 2 and 3 members.
The ESXi NFS client always uses the first port, also called vmnic0, for the network traffic. This is important to remember when you have hybrid 1Gigabit Ethernet/10 Gigabit Ethernet network configurations.
In a VSA cluster, the network traffic is divided into front-end and back-end traffic.
Front-end network traffic
Communication between each VSA cluster member and the VSA Manager
Communication between ESXi and the VSA volumes
Communication between each VSA cluster member and the VSA cluster service in a two node VSA cluster
vMotion traffic between the second switch and the hosts
Back-end network traffic
Replication between a volume and its replica that resides on another host
Communication between each VSA member cluster in a three node VSA cluster
Logical Network Architecture
Each vSphere Storage Appliance has two virtual NICs: one handles the front-end traffic, and the other handles back-end traffic. The back-end virtual NIC has an IP address from a private subnet. The front-end virtual NIC can have up to 3 assigned IP addresses.
IP address for VSA management network
IP address of the exported NFS volume
IP address of the VSA cluster (assigned only when the VSA cluster member is elected as the cluster leader)
The IP address of the VSA cluster can move between VSA cluster members. It is assigned to the front-end virtual NIC of a VSA cluster member only when that VSA cluster member is elected as the cluster leader. If the cluster leader becomes unavailable, the VSA cluster IP address is assigned to another VSA cluster member that becomes the leader.
Two vSphere standard switches on each ESXi host isolate front-end and back-end traffic. The physical NIC ports act as uplinks for each vSphere standard switch so that each NIC handles either front-end or back-end traffic. The standard switches use ESXi NIC teaming to provide link failover.
The following illustration depicts the logical network of a VSA cluster member that is the leader in the VSA cluster. The logical network of other VSA cluster members is the same with the exception of the assigned VSA cluster IP address.