Considering More than Two Nodes

The availability defined in the introduction is simply

$$A = \frac{T_{\rm Up}}{T_{\rm Up} + T_{\rm Down}}$$ (2)

One would like simply to replace $T_{\rm Down}$ by $T_{\rm Recover}$ and have that be the new Availability. However, life isn't quite that simple. In an $N$ node cluster, the Availability $A_N$ is given by

$$\begin{aligned}A_N = &{T_{\rm Up}} \left( T_{\rm Up} + T_{\rm D... ...\left. T_{\rm Recover}(1-(1-A)^{N-1})\right)^{-1}\\ \end{aligned}$$

So if really high Availability values are important to you, more than two nodes becomes a requirement.

However, the most important aspect of more than two node support is the far more prosaic cluster operation scenario: as the number of services ($\equiv$ hierarchies) in your cluster increases, the desire to increase the computing power available to them usually dictates larger clusters with one or two services active per node.