Advice/Tutorials for creating an Architecture Repository

[dels]

Hello

I’m documenting the baseline and target architectures for my company, using EA and the Archimate diagrams I can create the visual representations.
However I’m a little lost with regards to capturing all the attributes in a meaningful way, and using my EA model as the Architecture Repository.
 
I believe I want to
•   capture the attributes of the  systems ( things like product, version, lifecycle etc ) as tagged values.
•   then generate documents/reports of tabular data detailing these attributes

I have seen a short/simple video on creating tagged values, but I am looking for more detailed tutorial on how to create and report on sets of data for an Architecture Repository.

* Additionally, I’d really appreciate any advice or tutorials showing the best practices for setting up the model hierarchy for Architecture Repository.

Thanks kindly
Dels

[Glassboy]

Hi dels. 

I think partially you're asking about package structure.  If you're wedded to using Archimate then check out the TOGAF content meta model and use it as the basis for how you structure your content.  But keep in mind you need to separate archetypes from deployed elements.

Personally I also wouldn't put everything into tagged values.  For example I'd create an archetype system software "Microsoft SQL Server" and then where you're deploying that technology into a view of a system have it realize the archetype.  You can have version and patch level as tagged values on the deployed elements.  What realizing archetypes gives you is the ability to use EA's tracability window to quickly see where the same component is deployed across  as well as being able to do the same in reports.

Other than that it's a bit hard to suggest things without seeing how you use Archimate.

[dels]

Hello

Thanks for the response.

Are you saying that you'd model the generic products using Archimate like this :


Then the actual deployed objects like this ?


Dels

[Glassboy]

Hello

Thanks for the response.

Are you saying that you'd model the generic products using Archimate like this :

</quote>

No, not quite, because you're mixing behavioural and structural and the derived relationship between system software and infrastructure service is confusing.

If you want to model both the behaviour and the structural it's best to make everything explicit.  So I'd have the system software Microsoft SQL Server deployed on a generic SQL server node with the associated infrastructure functions and infrastructure interfaces.

This allows you evaluate the suitability of design choices by looking down from your Database Services infrastructure service to the infrastructure functions, and check the completeness of the design by looking whether all of the interfaces have been reflected properly in a deployment view.

<quote>
Then the actual deployed objects like this ?

No, because you can't instance elements in Archimate.  That's why you have no shapescript working.

I'd either have [deployed system software] --realises--> [archetype system software] or [server name (node)] --realises--> [generic node].  Or both.  It all comes down to what level you're documenting your environment to, and what reports you want to run.

[dels]

Hello Glassboy

Thanks for your response.

Folder Structure
------------------------
I'm new to EA, and while following the short tutorials helps get started with creating diagrams, I have a lot of questions with regards to how to be start a repository ensuring the folder structure is correct for my longer term needs.
As such I agree I do need some help / advice with regards to folder structure.
So far the folder strucutre I am working to is :


Behavioural / Structural
------------------------
In terms of showing both behavioural and structural, are you saying you'd model it like this ?


Archetypes
-------------
With regards to archetypes, I am a little unclear exactly what you mean - are you suggesting creating profiles / stereotype to create the specific product nodes/system softwares ?

If I simply have a two nodes, generic concept and and specific product on the same diagram ( component or archimate ) I don't have a realise connector :

Would you use specialise instead, or have I got the wrong diagram type / node  ?

Level of Modelling
------------------------
With regards to the level of modeling, what I am hoping to achieve is
1) document the type of components in our organisation - i.e. types of nodes, softwares, applications
2) document each deployed instance of these various components
3) be able to both a)produce diagrams shows at a high level/low detail architecture, as well as b)low level / high detail deployments.
4) be able to export specifics, such as version numbers, node names etc to tabular format through generated reports / documents


Thanks again, I appreciate your advise
Dels

[Glassboy]

Any chance you can fix the images so I don't have to fish the URL out of the page source?

This is an example of how I generally model the technology layer

[dels]

Hello

Thanks for that response.

With regards to the images, the image button only offers me square bracket pairs with img / img. The "Attachments and other options" has 4  checkboxes none of which pertain to images/attachments.
I'd gladly post another way, it would save me having to upload and share files on the internet.

Thanks for the image.

1) With regards to archetype versus deployed, I assume your example is showing the archetype components and no realised/deployed items.
If you were to show deployed components, would you show these on a non-archimate diagram, such as a standard component diagram like this (Deployment - Baseline Architecture ) ?


2) With regards to cataloging the version, names, ip addresses etc - again would you create a stereotype which was used on component diagrams and had nothing to do with the archimate section of the model ?

3) The folder structure in the image above, shows a high level branching between deployment ( specific items with catalogued details ) and archetype ( high level archimate layers with generic components ).
Does this appear to be a good way of achieving my aims ( documenting business level diagrams using archimate and cataloging all the technical details for our deployed footprint )

Thanks for the advice regarding the best approach / setup for my needs.

PS - I have just downloaded a swathe of the leanpub books as I gather those are the authoritative source on non-trivial use of Enterprise Architect, so hopefully I won't be asking what must seem like trivial questions for much longer !

[Geert Bellekens]

With regards to the images, the image button only offers me square bracket pairs with img / img. The "Attachments and other options" has 4  checkboxes none of which pertain to images/attachments.
I'd gladly post another way, it would save me having to upload and share files on the internet.

Unfortunately we cannot post images or attachments directly on this forum.
Most of us use a site like http://imgur.com/ to upload images and then use the img tag to show it here.

Geert

[qwerty]

The used links are fine. But this YABB obviously has trouble to pick the image. So for the Google drive you have to use the URL rather than the Image icon when posting.

q.

[Glassboy]

Hello

Thanks for that response.

With regards to the images, the image button only offers me square bracket pairs with img / img. The "Attachments and other options" has 4  checkboxes none of which pertain to images/attachments.
I'd gladly post another way, it would save me having to upload and share files on the internet.

Yeah google drive seems to serving them as some kind of blob, the forum software would probably deal with them fine as URLs.

Thanks for the image.

1) With regards to archetype versus deployed, I assume your example is showing the archetype components and no realised/deployed items.
If you were to show deployed components, would you show these on a non-archimate diagram, such as a standard component diagram like this (Deployment - Baseline Architecture ) ?

The example is one I created last year to demonstrate how the various information elements fit in, but the important thing in it from your perspective is the relationship between the technology service and the functions and interfaces.  The functions answer questions for enterprise architects (and potentially BAs) and the interfaces answer questions for solution architects.

As for the actual node, I have done something similar for Microsoft Exchange where I created Archimate nodes for each of the actual deployed servers and had them realise the generic node.  Each of them had tagged values with details like Exchange role, IP address, storage details et al.

I had similar generic views for SharePoint and a number of other technologies.

At the top of each package was a diagram which had a number of model views on it with SQL queries behind them.

One of these showed all the Exchange servers  (i.e nodes realising my generic node), one all the interfaces that Exchange presented that can used, and another showed all the technologies (generic nodes) that were using the exchange interfaces. 

I don't have access to that model any more, but the underlying package structure changed as I worked out more of the SQL necessary for the views.  There are a bunch of people on the forum tho' who are absolute wizards when it comes to that sort of thing tho'

[skiwi]

Try

Code: [Select]

[img]https://drive.google.com/uc?view&id=0B2s5WXrA04GJb1pUNDVYZ2E5OE0[/img]
rather than

Code: [Select]

[img]https://drive.google.com/file/d/0B2s5WXrA04GJb1pUNDVYZ2E5OE0/[/img]
to display the image

[qwerty]

Doesn't work either. It probably needs a fix in YABB.

q.

[Geert Bellekens]

Works for me. I can see the image posted by skiwi

Geert

[qwerty]

Funny. Now I see them too. When I answered I did not see them (directly). Strange.

q.

[Helmut Ortmann]

Hi,

i don't see it.

I assume it depends on the proxy / policies of your PC.

Regards,

Helmut