From what I see and understand of Quotas, they are not detected or provisioned by the differents services of OS.

So, for exemple, if I have 1 node for Cloud Controller, which exclusively aim to control and manage Compute/Volume nodes, hosting Glance/Quantum/keystone and Horizon services only, and let say, 1 node hosting Compute/Network Agent and Volume services only.

This last host is a Dell M610 Lame Server with 32Gb Ram, 2xIntel Xeon E5520, 500GB HDD (iSCSI Provided) and a root filesytem of 67GB which is not see by OS component.

If I go to the dashboard, default Quotas for my installation will be: RAM: 64GB HDD: 1TB vCPU: 20

Which seems to be hardcoded compared to the available hardware. So, let say once again that I create four projets as follow:

PRODUCTION / QUALIFICATION / DEVELOPMENT / SANDBOX

Those four projects come with the exact same default Quotas, but differents users, which are not aware of others. If every user create maximum ressources available instances at the same time I'll probably have errors throwing up because of the overcapaciting isn't it?

I mean don't would have been more logical to have a default Quotas detected and dependant of the actual ressources and then projects allowed Quotas dependant of the ressources allocated on the earlier projects?

I don't know if I'm crystal clear here and can reformulate my question if it's not enough.

PS: This question is not indeed a technical one but rather an opened one based on what I experiment on OS.