Modeling sub-service dependency

[rchevallier]

Hi all,

I'd like your feedback for modelling in archimate a subservice dependency
My understanding of archimate is you have several ways to model it.
For example for application layer:

For a simple api call dependency, you can use the serves relation: cf api-1.PNG

But you can model the request (as a trigger relation), and the answer (as a flow? relation): cf api-2.PNG
You can then also detail the data passed (parameters and answer)

My question: which one is the best? Should you model all of them ? Is the "serves" redundant or a derived relation is you model thru trigger/flow (request/response)?

[Jean-Baptiste Sarrodie]

Hi,

Both are perfectly valid and represent two different point of view. So none of them is "best", it all depends on your context and what you want (or have to) communicate.

Trigger and flows are dynamic and thus kinda focus on how things behave over time. I personally tend to use them when I have to discuss/communicate with application/software architects and developers to make it clear that some application "A" "calls" a service provided by application "B"

Serving is a pure dependency and provides no time-related informations, only the fact that something relies on something else. I tend to use it when I have to discuss/communicate with operations teams or system engineers to help them grasp impact of failure of application "B".

Regards,

JB

[rchevallier]

Thanks for your insight. It corresponds to my intuition.