Whose framing this is
The market-window frame is James L. Pelkey's, from the interviews behind A History of Computer Communications, 1968–1988, and is attributed to him here as in the two companion articles.
The trigger was the previous window's success
Local area networks worked, so organisations bought them — department by department, building by building, each choosing on its own. By the early 1980s a large company did not have a network problem; it had a network-plural problem: several islands, often on different media and different protocols, with no way across.
That is a trigger of the same kind as the one that opened the networking window: an accumulated consequence rather than a permission or an invention. Each window in this sequence is opened by the thing the previous one succeeded at.
What the window produced
Bridges first, because the problem looked like a wiring problem and a bridge is the answer that requires no thought from the endpoints. Then routers, once it became clear that a large enough bridged network is a single failure domain and that the interesting boundaries were logical rather than physical.
The companies founded in this window are the ones whose names still appear in enterprise price lists, and the ones whose acquisitions fill the late-1990s consolidation tables.
The standards war ran inside the window
This window is the one with an argument in it, and it is the reason the period is taught more than the others.
OSI had the institutions: international standardisation, national procurement mandates in the form of GOSIP, a conformance-testing consortium, and interoperability testbeds. had running code and a decade of operational experience, having replaced NCP on the in 1983.
The outcome is usually asserted. It is more useful to point at where it became visible: the two 1988 trade shows, months apart. The Enterprise Network Event showed an OSI ecosystem that was largely described rather than shipped; Interop put competing vendors' equipment on one live network in public, where failure would be seen.
One demonstrated a working network between rivals. The other demonstrated an intention. A buyer walking both floors in 1988 did not need the argument settled formally.
Why it closed
By 1989 the answer was known, and what followed was execution and consolidation rather than formation. The companies that were going to define internetworking had been founded; the ones founded afterwards were mostly bought.
The window closes while the market is still accelerating
Internetworking revenue was about $30 million in 1983. By 1988 it was $563 million, and still climbing steeply as the window shut — bridges and routers alone went from $20 million to $175 million across the same span.
Meanwhile the data communications sector that the first window created had already turned: $1.65 billion in 1986, $1.54 billion by 1988. One market was falling while the one built on top of it multiplied.
The pattern holds across all three windows: the window is not when the market is largest — it is when the question is still open. Revenue in each of these markets peaked long after the window that created them had shut.
Figures from Pelkey, Appendix A.1, Product Revenues 1970-1988.