Discovering Windows NT Server

The foundation had to begin with the security system in the network operating system (NOS) on top of which all applications would ride.  That wasn't going to be Netware.  

Apple had pioneered the building of computers that gave life to Douglas Engelbart's vision of what computers should look like.  That vision in turn sprang from an article by Vannever Bush that had been published way back in the July 1945 issue of The Atlantic Monthly.  

Engelbart, had given what came to be called "the mother of all demos" on December 9, 1968.  Using primitive computers and whatever presentation technology and trickery was available at the time, he showed the world the beginnings of what we now think of as the standard computer user interface. Apple got their first with their Lisa project, but virtually everything Apple ever built was in its own closed little world.  Integrating with the enormous variety the whole universe has just never been their thing.  But, it was clear that most computer interfaces that were used by people had to become graphical in nature pretty much the way Doug Engelbart had demonstrated.  Something like Microsoft's Windows was the way to go for most of the desktops and early portable computers.  However, that did not necessarily mean that something from Microsoft would provide the NOS.

In the summer of 1992 Novell's product had about 80% of the LAN server market, and they had three fatal  problems.  They had poor development tools.  They kept breaking applications that ran on their operating system.  And, they really didn't care.

 

The only development environment that Novell supported was an arcane version of the "C" programming language by Watcom that wasn't used by a measurable percentage of the world's software developers.  There were two reasons for that.  First, it was C, and not a very good version of C at that.  And second, Novell's engineers made absolute war on anyone that was trying to make a living building applications to run on their servers, and any customer brave enough to buy one of those applications.  One could go broke trying to build and sell applications that ran on Netware.

 

The interface between software and the operating system or another bit of software is called the application programming interface or API (very original that).  And every time Novell shipped even a trivially minor upgrade to Netware in the 1980s and early 90s, the broke their API.  If you updated your servers to a new release of Netware, your installed applications all stopped working.  So, you couldn't do an update until all your app providers updated theirs.  And then you had to find a long chunk of downtime when none of your users would mind having the servers down.  I had spent many a  weekend starting late on Saturday and lasting well into Sunday afternoon doing Netware updates.  I had visited Novell's offices and complained bitterly about it, and they just weren't interested in those complaints.  They were Novell, and like IBM, taking customer input was just not something they felt they had to do.  Like IBM, getting input from a customer was more about having some execs go play a round of golf together.  There was no way Novell could be trusted as our NOS supplier going forward.

Two years earlier, when IBM and Microsoft got divorced, the projects they had underway were divided between the two companies.  The joint IBM/Microsoft NOS, OS/2 LAN Manager in its 1.x version, was actually far worse than Novell's Netware.  The 2.x version of OS/2 was being developed on an obsolete chip set that Intel was moving away from as fast as they could.  But, IBM wanted the versions of OS/2 (desktop and server) that were furthest along in development, and Bill Gates was glad to let IBM have them.  He got to keep everything related to an early stage product which had been planned to run on Intel's latest chips, and which had been called the OS/2 3.x and Lan Manager 3.x projects to outside observers.  Also, he had hired the father of DEC's VMS operating system for their VAX computers, David Cutler. 

 

We didn't know it at the time, but Cutler had taken his unfinished source code for VMS 5.x with him to Microsoft.  But the thing that convinced me that we had to go with whatever it was they were going to come out with, which didn't even have a new name yet, was that in early 1992 before I had taken the AR&T job, Bill had given and interview in which he talked about this new operating system he was working on.  There would be a new NOS version, and he said he was going to give it the same API as the Windows desktop, with the addition of the functions required by a server.  And, (this is the important part) he was going to support it the same way he had been supporting DOS and Windows, complete with the same popular developer conferences, including the one in New Orleans which had become every developer's dream trip. 

 

Bill was going to turn every PC software developer into someone who could also build server based applications if that struck their fancy.  Even though this product didn't have a name and was more than a year away, it was going to give Microsoft the ability to take virtually the entire market away from IBM and Novell, and there wasn't anything they could do about it.  They had missed their chance.

I explained all of this to Joe and told him we had to find someone who could get us inside that part of Microsoft and make sure what they were building would work for us.  A few days later I had a meeting with Maureen Baskin, the sales rep assigned to the Boeing account.  It was in a park in downtown Belleview.  And I explained that I needed her help getting the whole of Boeing up on their new NOS, whatever it was going to be.  Surprised was not the half of it, in terms of how she felt.

It turned out that Microsoft would be giving very early beta copies of this new NOS to a lot of people to try out, and provide feedback into the development team.  Boeing already had several dozen early builds of it scattered around in various engineering labs across the United States.  So she got me a list of all of those folks, and I contacted them.  I explained my plan to build a single unified NOS spanning all of the LANs in the company and that I needed their help.  Every one of them was on board and excited to be a part of it.  And that was the beginning of what became the NT underground in Boeing.  We had a corporate-wide system up and running even before Microsoft officially changed the name of the product to Windows NT Advanced Server 3.1.  The BCS execs inside Boeing who were buddy-buddy with IBM were fit to be tied.  Orders were sent down through all of the layers of management, and Joe called me into his office and said that he had been instructed to get the list of all of the NT underground participants from me.  I had anticipated this.  I had deleted everything from my work computers, and was keeping all of my contact data at home.  I told him all of that, and he said good.  As I left his office he said: "You're doing a good job.  Keep going."  Which I did.

 

The NT underground I had put in place was unstoppable.  Boeing was getting it's unified NOS-powered network because there was literally an army of technical people building it, and their budgets and management chains were nicely spread across the whole company, and not identified as such.  It was the bottom of the pyramid all pulling in one direction.  All the clueless leadership of BCS could do was sit back, watch it, and cry.  They went through the motions of doing an "official" NOS selection process, but the results had already been dictated by what was happening on the ground.

Here it is worth saying something about the politics of IT in Boeing at that time.  There were two computing chief architects councils.  One was in BAC and the other in BCA.   Each functional organization, and the major full scale engineering development programs had a  management representative, who was ranked as a director or higher.  They were not all equal.  Typically only three or four mattered when it came to the councils making a decision.  These included the IT directors for the finance and engineering organizations, and those from any large programs currently in their development phases.  The company was very much oriented toward giving the big programs of the moment whatever they needed to be successful, at least it was before the merger with MD and the take-over by the disciples of Jack Welch's "Wall Street and the CEO first" nonsense.  What this meant for my work was that I didn't need the permission and support of the folks from whom Joe Muldoon was protecting me.  What I did need was the support of those key directors on the two computing councils.  They were my customers.  Making sure that I was always working for their best interests, and that they  knew that was my orientation was critical.  Convincing them that the leadership of BCS did not think that way was not something I had to worry about.  Also, the Commercial Airplane computing director for finance was personally, a hands-on participant in the NT underground, and easily had the most sophisticated early hardware setup for his servers.  The NOS approval was a slam dunk.

The tension between Boeing's IT organization (BCS) and the rest of the company was palpable.  One day back in 1989 when I was on the P-3 program, I was attending the program manager's staff meeting and a very interesting thing happened. The chief engineer, Jack Bloodworth had to report that he had just lost a couple of days of time on a critical piece of work.  Two computer technicians had been sent over by their BCS managers to reconfigure a couple of the program's VAX computers because they were not in the approved BCS configuration.  The VAXes were being used to simulate all of the mission avionics systems on the airplane, so the software engineers could develop the mission hardware's operating system, and the software that would run on it.  It was going to take a couple days to get the machines back to where they were correctly mimicking the plane.  Pete Downey, the program manager who would later be elevated to vice president of the new Navy Systems Division, exploded (Pete did that on a regular basis).  He got all red in the face and screamed: "I hate BCS!" and then stormed out of the room.  It would not be the last time I would hear an executive or senior manager say something disparaging about BCS.

 

As I would learn from people working in many other large companies over the next couple decades, the condition of IT departments not being seen as being as helpful as their teammates thought they should be was a fairly widespread phenomenon.  It was that insight that led to my first book, "The Tao of IT."   You can find out more about that book and how to get a copy elsewhere on this site.  The key takeaway here is that it was becoming widely known and appreciated (except by my own management chain) that I was effectively directing computing resources to help the two operating arms of the company.

Email was next.