The one place where the slides show Operations involved is in
emergency shutdown. But there isn't much there, OMIS should flesh
that out. I'd think we (OMIS) should make a more-than-one-slide
version of the Mini Use Case to explore it.
A number of questions arise. What policies does the NOC use to decide
to shutdown? What do they need to look at to get the evidence that
it's really a specific slice that's at fault? What else needs to be
considered? Does the NOC have any less drastic recourse?
What about handling other emergencies? What if a compute cluster gets
turned into a botnet and unallocated slivers actually start to
misbehave? If they aren't part of a slice there's no slice to
"emergency shutdown". If the compute cluster is just cut off, you
penalize experiments that aren't misbehaving but just happen to have
slice resources there.
-MAP
_______________________________________________
omis-wg mailing list
omis-wg@...
http://lists.geni.net/mailman/listinfo/omis-wg
