Author Topic: Product architecture divided into "components#  (Read 1506 times)

Mats Gejnevall

  • EA User
  • **
  • Posts: 92
  • Karma: +1/-0
    • View Profile
Product architecture divided into "components#
« on: March 19, 2019, 07:12:30 pm »
Hi all,
We like to divide our Product architecture in a set of sub architectures (Components) that are freestanding. When a customer ask for a variant of our Product we want to be able to merge these Components into an specific Project architecture for that customer.

In a perfect world we also would like that all users of a component are notified when a "Component" is updated in some way and then update the Component in the customer Product architecture.

In the same perfect world there are relations between the Components that magically are maintained and reintroducted when the Components are merged into a new customer Product architecture.

We tried the patterns function in EA, but it is very hard to use and it does not keep the notes pages on elements between releases of a pattern.

We could have one main standard Component that maintains the relations between Components??

Any ideas?

Thanks  :)
Mats

Geert Bellekens

  • EA Guru
  • *****
  • Posts: 12340
  • Karma: +489/-33
  • Make EA work for YOU!
    • View Profile
    • Enterprise Architect Consultant and Value Added Reseller
Re: Product architecture divided into "components#
« Reply #1 on: March 19, 2019, 07:19:39 pm »
Mats,

I did not understand what you mean, but I do now that "merge" and "model" do not go together. It basically can't be done.

Geert

qwerty

  • EA Guru
  • *****
  • Posts: 13142
  • Karma: +376/-299
  • I'm no guru at all
    • View Profile
Re: Product architecture divided into "components#
« Reply #2 on: March 19, 2019, 08:14:15 pm »
Also just at a guess: if you cut your model the right way you can have controlled packages holding those variants and you could assemble parts "on the fly" from these packages like Lego. However, creating variants isn't something that can be achieved easily. You probably have to find some (finally not elegant) compromise.

q.