Hi List !
I pretty new to archimate and i facing to a doubt on how to model, in brief i've 2 node with both a System Software As "Red Hat Linux".
My question should I model both nodes with only one System Software shared
(http://forum.archimatetool.com/index.php?action=dlattach;topic=514.0;attach=448)
or each node should have a unique System Software box (Red Hat Linux ),
(http://forum.archimatetool.com/index.php?action=dlattach;topic=514.0;attach=446)
but in this case I got warning from the validator as duplicate name entry and I loose from the visualizer the possiblity to query all the node that runs on RedHat Software
Thanks in avance for your advise
charbo
Hi Charbo,
It really depends what you wish to model:
- if you with to model the operating system (brand and perhaps version), then you may model a single System Software. This could be interpreted as: Red Had Linux is installed on both servers
- if you wish to model operating system instances, then you may model one System Software per server. This could be then interpreted as: Each server has bot its own instance of Red Hat Linux
So, it really depends on what you wish to represent in your model. All is a question of point of view. As it has already been said: models are wrong, but they might be useful ;)
If this can help you, I personally represent a single System Software. In fact, I use my database plugin (https://github.com/archi-contribs/database-plugin/tree/master/v2 (https://github.com/archi-contribs/database-plugin/tree/master/v2)) to create a repository containing all my SBB (Solution Building Block) and then share them across all my models.
This way I'm able to export my elements and relationships to a Neo4J database to do impact analysis.
Best regards,
Hervé
Hi,
As pointed out by Herve, there is no best choice. It all depends on what you want to communicate.
ArchiMate is an architecture modelling language and as such is made to create architecture descriptions. So this is all about communication: for who are you modelling? What are the concerns of those people? It's only with answers to these questions that you will be able to do a choice. For example, in most of my technical views, OS and Devices are almost never modelled and I just use properties on Nodes to keep track of this information. This allows me to keep this information without having to add elements and relations. This is very similar to having only one generic OS or Device that all nodes compose. But if your model is shared with people from operation team, you might have to add properties to keep track of exact version of OS, its languagem its architecture, etc. In such case, you are no more modelling a class of device but an instance and you'll then have to model one per node.
Regards,
JB
Thank to All for your replies !