The story of my involvement with IBM's EA

I have often mentioned on LinkedIn discussions and elsewhere my involvement with IBM's internal Enterprise Architecture (EA) function, but I suspect not in enough detail to clarify what I am trying to convey.  I think I should record the story here so that I can just post links over here when the subject comes up.

When I say "IBM's internal EA" I am specifically referring to the EA function that reports at a high level in IBM and has done heavy lifting in IBM's business transformations over the years -- including streamlining of business processes across the many regional offices and operations of a truly world-wide enterprise.  When I say "IBM's internal EA", I am intentionally excluding the enterprise architecture practices that provide enterprise architecture consulting services to IBM's services clients.  IBM does not necessarily follow the advice of the external service practitioners, and the internal work is quite advanced, with a history that goes back a couple of decades.

I date my involvement with IBM's internal EA effort from the late 1980s, before I actually became an IBM employee.  At that time I was a couple of years into my tenure at Pacific Bell, where in the role of data architect, I had assumed responsibility for an information and functional model at the enterprise level.  This "functional business model" had been created by a large team (100+ person-years) before I joined the company.  This was shortly after the AT&T divestiture, so the Baby Bells were feeling their way into the status of independent and competitive businesses they had recently acquired (as opposed to the former status of monopoly utilities).  The term "business process reengineering" was in the air.  I directed the efforts of the functional business model and data architecture team into detailed process modeling of several business areas (Operator Services, Centrex Provisioning, etc.)  We were using an early version of swim-lane modeling, and eventually covered a lot of walls of the HQ building in San Ramon with process maps.  A typical map would include a dozen work groups and similar number of applications, in a way that accounted for paper and data flows through the organization.  Such maps were popular with business users, who saw their work laid out graphically, emphasizing the interactions among the groups, sometimes involving IT, and sometimes not.

We used simple drawing tools, and experimented with AI-based simulations.  There was a swim-lane tool built on a Sun workstation (by Steve Johnson's team), that allowed swim-lanes to be hidden or moved, so that group-to-group interactions could be highlighted.  My personal love child was a semantic network tool that intersected process models with business language ontologies, built on top of some experimental IBM software.  This is documented here   This tool attracted the attention of IBM, through the interest of one of IBM's systems engineers on the Pacific Bell site (Jerry Bruemmer).  Our work came to the attention of a fairly recent organization in Armonk (HQ for IBM).  This organization was chartered to put a process management structure in place for the entire IBM Corporation, to help tame the complexity that had developed through the semi-autonomous geography-based IBM organizations around the world.  Around two dozen process executives were appointed and assumed responsibility for business processes that were intended to be standardized across geographies.

The head (Mark Langman) of that business process management unit reported one or two levels below the CEO, and maintained that position as the CEO changed to Lou Gerstner.  Mark visited my lab, and then followed up by sending a small team to work with us on the process modeling approach.  We had some close collaboration (under mutual nondisclosure) on tools and approaches that went beyond the CASE tools available at the time (Knowledeware, e.g.).  In my opinion, this was an early form of EA, although that term had not been invented yet.

Some water flowed under the bridge, while I took a detour as an independent consultant and then a practice head for a small consulting company.  Skipping over that, I eventually joined IBM in 1993, as a kind of business-modeling maverick within the new and expanding Object Technology Practice (OTP) in IBM's service business.  I made my living in this framework via client work on numerous engagements in various industries.  I also served on a team that took IBM's consulting methodology in a "workproduct-based" direction - by leading the articulation of methods in what we called the "business domain".  

Interspersed with that work, I also worked with Mark Langman's team, which had fully implemented the process executive framework, and had adopted an EA approach with an architecture review board (under Catherine Winters) that was managing the transition from independent business units to consistent process definition across the enterprise.  Our OTP groiup constructed an object-based business modeling approach, but more specifically I worked with Jim Salmons and others to webify the library of architectural documents that were reviewed by Catherine's ARB.

This all gave way to a corporate EA tool, based on Qualiware, with its repository basis and excellent extensibility.  From Paris, the leader of this tool initiative (Pascal Negros) created the most powerful EA tool I have seen.  The tool became the basis for governance as part of IBM's internal business transformation efforts.  It eventually grew to a website, open to all IBMers, with graphic depictions of rolling current, 1-year, 3-year, and 5-year views of process reengineering and supporting IT constructs.  There were hundreds of graphics, generated from the Qualiware repository and tweaked by hand.  These graphics were active, such that every icon on the picture was a clickable link into either more detail graphs, or detailed descriptions of every node and edge at the leaf level.  

One interesting episode occurred long after Mark Langman retirde, and the Enterprise Architecture team was headed up by Michael Martine.  Under Michael's direction several of us brainstormed and then spent some serious time working on the idea of 3D EA in virtual worlds, like Second Life.  We did some conceptual designs, and prototypes.  This was in intersection for me of "The story of my life in virtual worlds", which in turn deserves some work, as I write this in early 2013.

When I was on my last big engagement at IBM (a 4-month stint as chief architect on a 300-person project at Huawei, in Shenzhen) I had the privilege of working with a talented technical architect (Ken Chen) who appreciated the IBM EA database so much that he extracted the whole thing (all 600+ meg) onto a local directory which could be used to walk clients through the amazing functionality, and governable content.  When I say governable I mean that in IBM projects presented their business transformation projects and IT architecture in terms of snippets of architectural models, which were evaluated by the ARB at several gates in the project life-cycle (Concept, Design, etc.)

I fear that this exemplary EA program may have deteriorated in the years since I left IBM.  I know there was an intent to migrate to IBM's commercial EA tool, away from Pascal's brilliant Qualiware-based tool.  I think some of the balance between business transformation and IT architecture may have broken down under waves of managers passing through and getting their career tickets punched.  But from my perspective, I have never seen a client or other EA program that was anywhere close to the IBM internal work, in terms of deep, effective tool support, and thorough and effective governance procedures, with a balance between business transformation and information technology issues.