Now, a lot of the time, our hosts and storage are co-located in the datacenter. So, if we start building topologies like this we are taking up a lot of switch ports for ISLs.
A lot of time the ISLs are wasted connectivity. We buy an
infrastructure to join nodes together, not just to join switches together. So, at this point, we look at collapsing our fabric topologies into a single director. We still have to remember that this is a single point of failure. The previous slide showed several independent switches interconnected via what are called Inter Switch Links (ISLs). However, the cabling becomes very complex and in this configuration it is easy to see why Directors are considered as a better alternative. Director switches also save on the number of inter switch links, leaving more ports available for devices to be connected to the switch, rather than just another switch. Different vendors have significantly different architectures in their Director Switches – and consequently have significantly different features and performance specifications.