Something about Computer Networks that really annoys me now (but didn't when I first read it) is its OSI bias. The use of OSI terminology and the OSI reference model is not a problem; what's not so fine is the attitude to alternatives. So we have such gems as
"Virtually the entire computer industry has now agreed to a series of International Standards for describing network architectures. These standards are known as the OSI Reference Model. In the near future, almost all other network architectures will disappear."and the discussion of standardization doesn't even mention RFCs or the IETF protocol process! Such a view of the world must have seemed rather problematic in 1989, and now seems almost ludicrous (at least in Australia; perhaps things are different in Europe). Computer Networks is still one of the best general texts on networking I know of, however, and it deserves to survive to a third edition, in which Tanenbaum will hopefully bring it more into line with reality in this regard.
[ Note: The third and now fourth editions of Computer Networks have addressed most of my criticisms in this review.]