[services-wg] [cfwg] a proposal for slice stitching
Jeff Chase
chase at cs.duke.edu
Wed Feb 24 16:16:45 EST 2010
Robert P Ricci wrote:
> ...
>
> I like the fact that the SM has no special privileges in your design -
> would it be fair to say that it's more a service than a 'core' part of
> the control framework?
>
>
On reflection, this question strikes me as revealing somehow.
There is no part of the ORCA design that has any special privilege.
Every actor speaks for some stakeholder: the AMs speak for the
infrastructure aggregates they contribute, and the SMs speak for the
users (experimenters), or rather for their applications
(experiments). Brokers have only the ticket-issuing rights delegated
to them by the AMs that choose to federate through them, and the tickets
they issue don't carry any special power if the AMs choose to
over-commit themselves or renege on their promises. Trust is anchored
in identity providers outside of ORCA, and each actor controls its own
policies regarding what powers to give to identities endorsed by those
providers.
So by this definition there is no 'core' part of the ORCA control
framework. In the case of GENI there is a suggestion that we might
constrain the actor graph in various, e.g., by asking all contributing
AMs to federate through a common broker that acts as the GENI
"clearinghouse". But that's a deployment choice rather than anything
inherent in the control framework.
This is why when anyone shows me a box-and-arrow diagram, I am always
pointing at the boxes and asking who each box speaks for and runs for.
I am genuinely confused by entities like Slice Authority in the SFA,
which seems to have some exogenous power to bless slices, or Management
Authority, which seems to have no power beyond that of the aggregates it
"speaks through".
Jeff
More information about the services-wg
mailing list