Ad RSpec: Include external references

[ahelsing@bbn.com]

The Ad RSpec should include all external touchpoints: places where ExoGENI resources connect to I2, NLR, Starlight, or other networks, and through which an Experimenter might stitch other GENI resources to ExoGENI resources.

[ibaldin@renci.org]

See my notes in #25 - currently nothing is 'external'. We may expose individual network AMs to doing simple stitching operations rather than doing entire paths as it is today, but they will remain under the control of a single a ExoSM (or at the very least there will NOT be a SM-AM mapping the way it is done for racks). This means the ad coming from that SM will still contain multiple advertisements that may or may not have 'external' touchpoints. Until the full power of ORCA tickets is exposed via GENI AM API I don't believe this is fixable, since we will not expose stitching without tickets.

[ibaldin@renci.org]

I should add that if you do listResources on individual racks you should see information about where the racks is stitched to (the type of external advertisement you're looking for). The problem mainly concerns the ExoSM because it sees everything and cannot separate individual Orca AMs repsonsible for various sites or network providers.

So this problem is specific to network Orca AMs like StarLight? and others, which are currently accessible only via ExoSM. Individual racks we should be able to make behave as required (with some tweaking).

[ahelsing@bbn.com]

If I understand your second comment, you can make this work for individual racks, just not the ExoSM? And per your comments on ticket #25, you'll add those external references at some point in the near future?

I'm much less concerned about network-only Orca AMs like Starlight.

I'm a little concerned by your first comment, that you won't expose stitching without Orca tickets. Maybe you can say more about what you mean here. My hope is that AM API v3, with its separation of Allocate() from Provision() might allay some of your concerns. But maybe that's not relevant given your later qualification that you can make the individual racks work as desired.

[ahelsing@bbn.com]

I've split this into tickets #38 and #39.

[ahelsing@bbn.com]

for clarity, close this in favor of 38/39