[Hawkular-dev] Inventory browsing UXD

[Thomas Heute]

We want to shift focus away from the "infrastructure focused" view to 
"application focused" view.
In that sense we'd rather traverse resources by relationships rather 
than plain parent/child like in RHQ.

The starting point should be an "application" a logical piece for the 
user. His Intranet, his e-commerce website, his b2b backend... (In fact 
I think we should let the user pinpoint the resource of his choice as 
starting points but if we can do for an "Application" we can easily do 
for a host or a server) (We'll fine-tune the notion of "application", 
OpenShift also has a related notion)

IMO an application is just a "virtual" resource with a collection of 
resources and some metadata, for instance a load balancer + EAP domain 
(and its EAP servers) + Database.
That resource will have a KPI associated with it and other information. 
It could also just be an "ear" file.

In terms of visualization, we would certainly like to see those 3 pieces 
(in that simple example) and be able to "zoom in" into one of them (to 
configure the load balancer), to see database metrics or to see the EAP 
servers and domain controller that forms the EAP domain.
The user should also be able to "zoom out", from a deployed ear to the 
EAP domain it's deployed on...
The user should also be able to look at the infrastructure (is EAP in 
some docker container ? how is the CPU/memory ?)

We'll also want to reflect health in those relationship, if my database 
is down, my "application" should reflect this.

It's going to be a challenge to get this right, with the right level of 
granularity/detail, but the user experience should be much better in 
most common situations.

For this we also need a good understanding of the relationship between 
all the resources, it should be mostly automated.

Thomas