In COST-2819, we created fake OpenShift projects (Platform unallocated, Worker unallocated, Platform) to quantify the cost of the control plane and the cost of the unallocated capacity. This works well and fulfills the need where users are looking at cost per-project because they are thinking in CMDB-mode: "projects X, Y and Z make up application A, so I'm taking cost of X, Y, Z plus the fair share of the control plane, plus the fair share of the unallocated capacity and that's the cost of my application A".

Increasingly, users are tagging their projects (all their resources, in general) to determine e. g. "everything involved in application A is going to be tagged <application=A>", or "everything I need to bill to cost center 1234 will be tagged <cost_center=1234>", or... In that case, we are not providing the cost of the platform, worker unallocated or platform unallocated.

Let's fulfill that case!

Same as we created fake projects for the Group By Project perspective:

• We should create fake tag values for the Group By Tag perspective
• No matter what tag key is selected, fake "Worker unallocated", "Platform unallocated", "Platform" (if the "sum platform costs" toggle button which we will need to add too is enabled) tags should be created
• This should be available wherever Group By Tag is possible (OpenShift page, Cost Explorer page, exports, API, etc

During the face to face we discussed grouping by the tag in the distribution so that we can keep the user projects tag values. However, we may need to limit the distribution keys if we run into performance issues.