The following text is copyright 1995 by Network World, permission is hearby given for reproduction, as long as attribution is given and this notice is included.
By: Scott Bradner
William Johnson from IBM was quoted in an interview published last week (NW May 8 1995, and on the web with a URL of http://www.nww.com/ibmqa.html) in answer to a question about the customer confusion level over IBM's TCP/IP Vs. APPN strategy: He said "We tell people and we tell the sales reps to find out what the customer has and if they've made a commitment, stay with that commitment, if it's TCP/IP go with it, if it's APPN, go with it. There's no reason to go in with every new product and try and get them to move." (I do like the way he differentiates between sales reps and people.)
This is good news if it is real. I do not doubt in any way Mr. Johnson's sincerity, but as Don Haile said in response to the same question "Its hard for our sales reps too."
I've been watching the internecine warfare at IBM over this topic for a while now. For the past few years there have been two rather separate camps within IBM on the true path to networking nirvana. The old-line SNA people who were embracing APPN with one level of enthusiasm or another, and the TCP/IP group. There were purists on both sides who insisted that anyone who traveled the other road was crazy or a bit worse, even somehow disloyal. There were many more people somewhere in the spectrum between these extremes who felt that both protocols had a place in the IBM portfolio. But the extreme views seemed to dominate the sales pitch in too many cases. This seemed to be more often the case with the anti-TCP/IP faction than the pro.
I watched a presentation by an IBM speaker at the SHARE conference that consisted almost entirely of denunciations of TCP/IP and listing reasons why it could never be useful. No solution to the customers' needs was forthcoming, just a litany of problems. A side note, the symbol of the TCP/IP group at SHARE is the bumble bee. This symbol is in homage to the old saying that aerodynamically a bumble bee can't fly but don't bother telling it that. Bumble bees and TCP/IP seem to be doing quite well in the face of apparent scientific improbability.
Last year I went down to the IBM center in Research Triangle Park North Carolina to give a talk about TCP/IP and the Internet as part of an attempt to get more of IBM responding to customer requests rather than telling the customer what was best for them. It was an interesting experience and I had a good time although I did spend the afternoon with some people who seemed to want to get me to see the true way represented by APPN.
APPN and TCP/IP are two very different protocols. Neither is the best solution to every application. It would seem to me that organizations that have been a pure SNA shop and who will continue to be, in the vernacular, all blue, should seriously consider APPN especially after High Performance Routing (HPR) becomes widely available. It is just as clear to me that a company with little or no SNA and with a deep involvement in the Internet and TCP/IP and has a heterogeneous hardware mix, would find a switch to APPN very difficult, incompatible with the Internet and thus not the right thing to do. The question is what to do in the range of situations between these extremes.
If Johnson & Haile can get the IBM sales force (all of those I have met are people) to explore these inbetween cases in an unbiased way with the customers, it will be far better for IBM and for the data networking community. I wish them luck, but from what I have seen they will also need to show strong leadership and a willingness to slap around some of the long timers.
Disclaimer: Harvard does not do columns other than those with a capital, so these must be my opinions.