Architecture-Based Transformation of a Large Enterprise
Hans Gyllstrom, Enterprise Design Group
This talk presented observations and insights gleaned from leading transformations to a better state over the past 10 years or more.
Case study of transformation of a large Australian bank. CEO wanted to be able to say: I’ll never buy another data center, rack, server, storage, or network device; never get locked into proprietary hardware; tell teams it will be weeks to get them new hardware; pay up-front for infrastructure that may not be used; implement internal solutions for common problems when an equivalent commodity cloud solution is available. Implemented an IT agile operating model.
Transformation: from current state to target state via a migration path involving numerous intermediate state.
Architecture comprises decisions on structure. Period.
Goal modeling is a better language for dialoguing with business and non-IT people.
To break down high-level goals into subgoals is damn hard work, and hard to get right. Doing so requires mapping the entire enterprise, with tool support, to goals and subgoals.
Key architectural learning: It’s more about platforms than resources. Services are subsumed by the notion of platforms. Evolving architecture of platforms above infrastructure.
Enterprise architecture fails because the scope is too broad; it doesn’t fit with how organizations are structured. People try to reuse architectures across business units. What is needed is platform-oriented enterprise architecture, not architecture frameworks.
- software people in large enterprises are unafraid to take on any complex problem, but they are too busy and so the rest of the universe passes them by; they don’t want to learn, yet they are anxious to take huge risks.
- architects don’t know how to architect–commoditization, PowerPoint language only, secondary effects very costly. Everybody and their brother is an architect. on one project cited: 120 solution architects, and they don’t know how to do UML modeling.
Ethnomethodology - Ethnomethodology’s Program, Garfinkel, Rawls, & Rawls: “If knowledge is knowing how to do something, then knowledge cannot be acquired by the written or spoken word.” Knowledge can be acquired only by working on the shop floor.
Organization and Control
People use control to mask lack of competence. Partnerships with vendors are difficult because procurement wants to beat the s**t out of them.
Left-to-right process leads to messes: plan–>design–>build–>run. Architects and “run” (operations) are the only ones who care about doing things right. Ultimate control may be with the run group. Car industry has addressed this problem with “design for manufacturing”: Create a set of constraints to which manufacturing responds, then follow a right-to-left process. For example, car must be these colors, this body weight, get this kind of gas mileage, and so on–begin with these kinds of run constraints.
People Politics and Culture
- use of organizational hierarchy and communication control to “manage the message”
- dictatorial-style managers with all the answers
- learning and unlearning not rewards
- not knowing who is friend and foe
- Choice Theory: A New Psychology of Personal Freedom, Glasser
- Mindsets: Fixed and Growth. Mindsets: New Psychology of Success, Dweck: Fixed – believing that basic qualities are carved in stone.Urgency to prove yourself over and over; live in world where some people and superior and some inferior. Growth – believing that your basic qualities can be cultivated through your own efforts; belief that true potential is unknown.
- What our words say – Precise words you use to communicate reveal more about you than you can imagine. They reflect psychological state rather than what influences or causes it (see The Secret Life of Pronouns: What Our Words Say About Us, James W. Pennebaker). Certain words predict honesty and deception.
- When you create something, the opposite is created
- When you hit an obstacle, be like water and flow around it
- The harder you push, the more resistance you get
- You may have to declare break-down in order to get break-through
Conclusion: competence, transparency on facts, goals, and assumptions; never question intent