Objective 6.4 – Describe the two-node cluster Architecture

In this post we are going to continue on and talk about a VSAN Two-Node Cluster Architecture looks like.

A 2-node cluster consists of two physical nodes in the same location. These hosts are usually connected to the same network switch or connected directly using network crossover cables. Use of crossover cables eliminates the need to procure and manage an expensive network switch for a 2-node cluster, which lowers costs—especially in scenarios such as remote office deployments.

A third host called a “witness host” is required for a 2-node configuration to avoid “split-brain” issues when network connectivity is lost between the two physical nodes. A witness host is a virtual appliance provided by VMware that runs ESXi. We will discuss the witness host virtual appliance in more detail shortly.

Two-Node with Direct Connect

• Connect two nodes directly between 10G NICs:
  • Remove physical switch requirements for vSAN Data (save $)
  • Use 2 direct connect connections for higher network availability.

Characteristics of Two-Node Cluster

• Built on the foundation of fault domains
• When using two-node clusters:
  • Both hosts in remote office store data
  • Witness is central office or other site
  • 500 ms RTT latency, 1.5 Mpbs bandwidth from remote site to witness
• Can scale to more nodes if needed
• All sites can be centrally managed by one vCenter.

Use Cases for Two-Node Cluster

• Two-Node architecture is ideal for remote office or branch office scenarios
• Up to 25 VMs per site with Remote Office/Branch Licensing (at time of this post).

Source: VMware vSAN & HCI Fundamentals & vSAN 6.6 Technical Overview