I might be missing something square in front of my face (or in my template or understanding), but when attempting to update a "deployment", I have a few, well, WTF's - about this whole process...
Assuming I'm not missing something simple - a 'deployment' can contain any object type in vRA, so seemingly:
- multiple machines from multiple, different cloud templates
- add'l disk resources, attached/assigned to machine resources
- multiple networks assigned to machine objects/NSX related network objects (we aren't there yet in my env)
- etc.
Assuming I'm not misunderstanding the purpose of said deployment (and I'm comparing the concepts to your Azure's / AWS's and their general interface wizards to a certain degree where you can create anything and dynamically set it to either a new or existing deployment you are authorized to see)
My sticking point is this (as clearly as I can explain)
From an end user perspective, start of a new deployment:
- they hit the service broker, request a template from the catalog for a single machine, single network (no add'l disk for now - making this example as simple as I can)
- This prompts the initial creation of a deployment - "MyNewDeployment-App1"
- Single network object and machine gets deployed - all processes run, etc, etc - new deployment "MyNewDeployment-App1" now has these 2 objects (VM and Network it's connected to
Ok, perfect, makes sense. User selected values for the template and everything got tagged/steered/provisioned/ip'd/config'd, etc, etc, etc based on the values they selected.
Now, time passes, the app workload grows, etc, etc and the end user now wants to add a web server or a (web server cluster of 2 VMs) - it doesn't really matter, they just want to add something to "MyNewDeployment-App1"
How the hell is this supposed to be accomplished? I know of the Update action on a deployment, but upon clicking on that, I'm presented with a screen that has 2 side entries items - 1 Update and 2 Plan, however in the Update window - all of the fields that appear are all of the prefilled data that the end user submitted on the first deployment that spawned the first machine in MyNewDeployment-App1".
Aside from this not making sense to me (hence my question) - there is no selection as to what other type of template they may want to add into this deployment or it's unique values.
This leads me to believe that I can only ever have 1 type of template deployed in a deployment (not talking about onboarding either), and I hope that's not the case as it seems wildly limiting and if that were how it would be, some of our end users might have 100's or even 1000's of vRA Deployment to cover, say, the web cluster of app A, the middleware cluster of app A and the db cluster of app A, then multiply that all by the apps our end users manage - it's not very feasible, scalable or remotely end user friendly.
And I'm aware that I can build a "app" template that has all the preset/hardcoded items of a web server/middleware/db and set the 'allocatePerInstance: true' and set count and use count.index, but this isn't a very flexible option
I'm looking to see if I've missed some yaml value or just the design construct of vRA as a whole in this regard.
Can anyone explain this and/or what you've seen or done?
If a vRA deployment can hold "anything" and be added to over time other then just the initial template that was used to create it or is it literally just that single template = a deployment and that's it and really the only thing you can do is have set the count in the template and they can adjust the count to scale that same machine count up?
Thanks to everyone who reads, sorry for the long post
