I didn't read the GoF book until the early 2000s. I gave up on it after reading their recommendation to use the "state machine pattern" for network protocols- that was a popular approach in the late 70s and early 80s, but by 1990 or so everyone knew the ways in which that was a bad idea. (I worked with some of those state machine implementations in the 80s)
There's a tendency among people who think of themselves as experts in "Software Engineering" to think that domain area expertise is irrelevant. This book is an example of that view.
Do you mean the RDMA verbs API? Or something else? I'll point out that it's an API, so an implementation pattern isn't relevant. (it's also an awfully obscure one, and I say that as someone who was involved in standardizing the pre-Infiniband Virtual Interface Architecture API that it derives from)
It's a relic of the old 7-layer model, where you looked at each layer in isolation, and was ditched in high-performance networking code starting with Van Jacobsen's in maybe 1991. No one ever hit 10mbit/sec with a state machine-coded stack.
Note that very few people write network protocol implementations, so it baffles me why they thought to suggest that their high-level design pattern would be useful in low-level kernel code.
There's a tendency among people who think of themselves as experts in "Software Engineering" to think that domain area expertise is irrelevant. This book is an example of that view.