Last month, you successfully merged two (2) local SAN islands into a single core/edge fabric topology and connected storage devices to the core switches. Performance benchmarks were met or exceeded. I

Conflicts between SAN routing protocols are uncommon in a properly configured fabric and would typically manifest as connectivity failures or more severe, unpredictable issues, not just a gradual slowdown with increased I/O.

Storage port buffer credit limits would restrict I/O at the storage port itself, but the question states that storage ports are 'under-utilized,' meaning the bottleneck is occurring before the data reaches these ports.

Multiple process login failures would cause connectivity issues or I/O errors for specific devices, not a general performance slowdown across the fabric while storage ports remain under-utilized.

In a core/edge SAN topology, all traffic between edge switches and core switches passes through ISLs. If some ISLs become over-utilized, they will introduce latency and reduce throughput for I/O requests, even if the final storage ports have ample capacity to receive the data.