Greenfield projects are difficult to do well because they require predicting the future accurately over a long period of time. It is better to focus on designing systems that can evolve and change easily in response to feedback rather than attempting to build a complete solution from scratch. Modular designs that allow replacing parts independently and retirement planning can help systems adapt and last through multiple generations.
4. Long term views required
Greenfields cost time and investments
+
Greenfield opportunites are big and rate events
=
Must cover a large chunk of future
->
But the Future is harder and harder to predict
6. Illusion of Done
Goal is to delivery the complete thing
Versus
continuous improvable thing
7.
8. Danger big monoliths
Big integrated solutions
->
Less focus on modular & replaceable designs
->
Unknown dependencies
->
Stuck with the solution for a long time
Consider : Multiple generations
9. ____________________
< I don't like grass >
--------------------
/ ___ /
// / /
(( O O ))
/ //
/ | | /
| | | |
| | | |
| o |
| | | |
|m| |m|
10. Long Feedback Loops
Starting from scratch + big goals
->
Long time to deliver
->
Separates Dev from Ops/Users
->
Long time before feedback
(from users, reality, operations)
->
Lots of invested time and money
->
Hard to change course
11. _____________
< Happy now?? >
-------------
(__)
/oo|
(_"_)*+++++++++*
//I#I
I[I|I|||||I I `
I`I'///'' I I
I I I I
~ ~ ~ ~
Scowleton
12. Creates Legacy
What about the last generation greenfield?
&
What about the cows in the old fields..
->
Deny new features?
&&/||
Wait until they die/leave?
14. From Cradle to what??
Think about Retirement when designing
->
Even consider a test retirement
->
Easier to replace/upgrade parts when needed.
15. In summary – Design for Change
• Now is old and the future is fluid.
• Changeability is the main feature
• Usage and Fast Feedback is essential
• Run 1 system with multiple generations
• Consider obsolescence from start
(Note Greenfields are not forbidden.. Just hard
to do right)