IBM really missed the boat with the new generation of networks. Once upon a time IBM dominated wide area data networking with SNA, and were making a play for the local area with Token Ring. At the same time the minicomputer and PC industry were developing the lower cost Ethernet and, courtesy of the Internet, TCP/IP and related products.
Quite correctly IBM pointed out that SNA was far more reliable and secure than IP. But it should have known that a major problem was brewing with this stance. The rigidity of SNA, ideal for centralised, host based transaction networks, wasn't suited for the mass of PCs connected on a Lan. Admittedly IBM reacted by introducing Netbeui protocols for the PC networks, but they wouldn't inter-work with host systems.
In contrast IP steadily matured, until in the hands of HP, Sun, et al, it was implemented right across the board, from wide area to local area networks. One protocol for all, including the communal network, the internet. As we all now know, the internet also enforced the standard, so that even Microsoft obey the rules!
Eventually IBM capitulated, and integrated Ethernet and IP into their whole product range, rather successfully in fact. But by then a number of specialist communication companies had developed product ranges to support the new generation of IP router based networking.
Because of details in the implementation of routing protocols and IP addressing handling, the protocols from different companies didn't work together as well as they should have, and so the market matured around two or three major players, of which Cisco is by far the most dominant. IBM tried and made some inroads with its own IP router products, claiming integration with mainframes, but even its alternative mainframe IP software products took a big market share.
And so the scene gradually reversed, with Cisco supporting IBM protocols on their IP products. The move towards Cisco escalated when it introduced direct S/390 channel-attached products. The industry had always been afraid of plug-compatible channel attached products of any sort, fearing that IBM could tweak the channel control code at any time. But the Cisco product was built on IBM approved software, the beginning of a relationship between IBM and Cisco, rather than the fierce head-to-head competition that preceded it.
It now looks like IBM has abandoned the IP networking products to Cisco, because the two companies are working very closely together. IBM these days is very successful in the systems arena, largely because it has adapted and responded to the inevitable need from customers' demands for support for a mixed set of systems.
Thus, while IBM obviously will favour Linux over NT and Aix over Solaris, it does offer comprehensive support for third party systems. Supporting NT, Solaris, etc, is no great problem for IBM. It has all the technical resources; NT is very similar to OS/2, Solaris to Aix, etc, so that support only implies a bit of extra training. But supporting large scale networks as part of a total system is a different proposition. And recognition of this fact has led to this close working relationship with Cisco.
Given IBM's long term focus of systems integration across the enterprise, a joint venture with Cisco makes a lot of sense for IBM. But what about Cisco? It is now (with a little help from a judge!) the highest capitalised IT company. It dominates the IP product market.
But competition is coming. Voice-over-IP (VoIP) is near to a reality, which means that the telecom network suppliers must be actively looking for a dominant role in the expanding internet to compensate for loss of conventional telephone revenue. New services, particularly wireless, will be the most critical, but while BT, et al have been content to work with Cisco at the periphery of their networks, they must be thinking of adding more value themselves.
As long as the current situation of adding data traffic over the voice networks exists it doesn't matter, but when voice and data are mixed over an IP network the situation needs rethinking. Thus the cooperation with IBM makes a lot of sense for Cisco as well as IBM. In effect Cisco has IBM on its side, not on their backs, so that they can concentrate on the up coming battles with the telephone industry.
These are battles I welcome, because we have seen all too clearly how competition has already improved services and brought down costs, and how lack of competition has strangled progress in office software.