2012-03-29 11:32:27 |
Eoghan Glynn |
description |
Currently quotas can be reduced to below the current usage level for that resource type, giving nagative headroom which seems to me non-sensical.
Was this intended as a deliberate way to "softly" reduce resource usage? i.e. to cause the in use count to slowly approach the new quotas as resources are released as part of their natural lifecycle, e.g. when instances are eventually delete'd, new instances cannot be spun up to replace them.
This however seems a very unreliable way of reducing reducing usage retrospectively, as the resources in question may be held onto indefinitely. A more reliable approach would be to force the release of the excess resource *before* the quota is reduced. |
Currently quotas can be reduced to below the current usage level for that resource type, giving negative headroom which seems to me non-sensical.
Was this intended as a deliberate way to "softly" reduce resource usage? i.e. to cause the in use count to slowly approach the new quotas as resources are released as part of their natural lifecycle, e.g. when instances are eventually delete'd, new instances cannot be spun up to replace them.
This however seems a very unreliable way of reducing reducing usage retrospectively, as the resources in question may be held onto indefinitely. A more reliable approach would be to force the release of the excess resource *before* the quota is reduced. |
|