Optical transport networks are at the turning point from providing semi-permanent connections to start being engineered for efficiently supporting dynamic optical connection services. Vital for these transition is the availability of node architectures that improve the flexibility of add/drop port utilization, thereby enabling the transponder connected to a port to be reused to realize connections with different wavelength channels or using different input/output fiber links (directions). This paper assesses the network-wide blocking performance provided by node architectures with different degrees of freedom for port reuse. The performance assessment not only clearly highlights the relevance of maximizing the chances of transponder sharing in transport networks, but also shows that the network topology determines which of the degrees of freedom for port reuse is more relevant.