Best way to handle transition and future state?

[pmason]

What is the best way to manage a transition and future state arechitectures in EA?
Do you create duplicate models, one for each phase/state?
Do you create a base set of components and use instances in diagrams?

Just playing with options, using the same component definition doesn't work as any link/assosiation you create are relected in all the diagrams.
Thanks,

[qwerty]

We just model one single time beam since we do not have the budget for multiple parallel projects as in The Deadline by Tom DeMarco. So we have one repository.

q.

[pmason]

It's not so much that we run parallel development, really we don't but there are a number of projects that contribute to the larger architecture.
The real reason for having the future state and transition state architectures is to show where we are going and how we are going to get there.

[Luis J. Lobo]

Take a look to the "GAP Analysis Matrix" introduced in EAv9. This will help you a bit:

http://www.sparxsystems.com/cgi-bin/search/search.pl?Terms=gap+analysis+matrix&Realm=EA+User+Guide

http://www.sparxsystems.com/enterprise_architect_user_guide/9.0/navigate_search_and_trace/gap_analysis_matrix.html

Regards,

Luis.

[pmason]

Luis, Thanks I will looking to utilizing the Gap Analysis tool, that would be the next logical step to see the differences/gaps.
Just skiming this it looks like you have to duplicate all the elements/components you don't create a common set and reference them in each model.
Thanks,

[Luis J. Lobo]

From a birds-eye, it could be several ways to do this:

Baselines comparision: 2 repos with the SAME elements (same GUID). 1 repo for "As-Is" models, 1 repo for "To-Be" models. Then, use the compare utility for see differences.

GAP Analysis: 1 repo with duplicated elements. GAP matrix helps to relate elements from "As-Is" vs "To-Be" and identify GAPs and tasks to convert "As-Is" in "To-Be".

Instances: 1 repo with 1 model with the Architecture components. Then, 2 models to instanciate the components of "As-Is" and "To-Be". Matrices and specific traceability diagrams will help to relate the instances of the two architectures. It will be a good option or not depending on the artifacts you're working with for architectural components modelling...

... and any other way that sparx or another user will propose to you!

[pmason]

Luis, Yes I was starting to come to those realizations. Even when I look at the TOGAF MDG its not clear, but it almost looks like it would either have to create copies of items from the continuum or instances but none of the TOGAF documents actually describe how to use it.
But thanks for you help, either going down the duplication or instance route. Having to deal with a number of regional architectures that we are trying to align with a single standard one. Hence there is some common global elements and then there are some regional elements.

[qwerty]

Actually we're doing a similar thing. We have many projects that (in future) shall influence a product. And vice versa the product will enable projects. We (currently) have organised it so that the product has r/o copies of existing projects and each project has a r/o copy of the product. Every now and then the r/o copies are updated. References are only allowed from project to product. Quite mind boggling. And still searching for the right way.

q.

[pmason]

Here is what I'm thinking, believe its pretty close to what Q. was describing his model structure is.
Basically the regional projects/models can drag elements in from the global product and common projects into their models. But each of the regional states, current, transition/phase, future are copies of each other which you would manually have to make sure if anything from a base/common changes get copied.

[qwerty]

Different things but comparable. What we also do is Require Lock to Edit. For the product->project view the project roots are locked by the administrator and vice versa for the projects the product copy is locked.

q.

[Darrell Piatt]

I have another low-tech way I've been handling it and wonder if someone has a better technique.  I basically copy a diagram and then add/remove elements to the next/previous phase.  You can elect to hide connectors in just that diagram.  You can also color-code to effect just that diagram.  This lets me do a little "time-lapse" photography in showing the evolution through the phases.

[pmason]

So I have been taking a snapshot of the whole model and c/p it into a new model and begin modifying. Like you are saying then I have the same diagram but what does it look in phase 1, phase 2, ... etc

I'm no expert at the tool as I'm still learning myself but the problem I would see with your approach is if you add new linkages/communications between components which are additions they would show up on all the other diagrams that have those two components. So you would have to go through all them to hide the new links/communications in the old diagram. Then again maybe there is some feature to prevent this. Initially I thought that I could just lock the diagram would prevent anything new showing up in that diagram but that is not the case when it came to links/communications between components that already exist in that old diagram.

[Darrell Piatt]

You're right about that.  On more than one occasion, I've hit this weeks later when I go into a diagram I haven't viewed in awhile.  EA also has the diagram filter feature, but that also has its pitfalls for this.

There are a lot of features and perhaps I'll figure it out at some point (or see it posted here).

[OpenIT Solutions]

Just a quick not on this one - there is a feature to lock a diagram - i pretty certain it would stop now relationships showing...

Also worth mentioning for those that haven't seen yet 9.2 has a diagramatic baseline compare feature. This combined with diagram filters for phases gives quite good capability in this space...