#opendaylight-group-policy: ODL-GBP-MODEL

Meeting summary

1. service_insertion (dconde, 17:08:34)
   1. we need to determine whether GBP does the plumbing for other service insertion projects (dconde, 17:09:29)
   2. readams says a pair of EP talking via a FW -- but we are restricted from dealing with this in a pairwise fashion. (dconde, 17:10:16)
   3. ways to reason about it. (dconde, 17:11:34)
   4. dvorkin says - group provides a contract … (he's drawing on board) then another group consumes a contract. (dconde, 17:11:52)
   5. within this grop, there is a subject to the conversation., let's say it's for HTTP for sake of argument (dconde, 17:12:11)
   6. action is specified to redirect to a service GRAPH (dconde, 17:12:22)
   7. let's actually call it a chain. (dconde, 17:12:31)
   8. there's an entry point an an out. and we stick devicess iside. for service functions. we may have direct return too. (dconde, 17:13:05)
   9. expressed from consumer's POV, we enter the service and we pop-out. It's VERY easy, right? (dconde, 17:13:28)
   10. but now, for the details. (dconde, 17:13:36)
   11. what if we try to do the following: (dconde, 17:13:46)
   12. we have two VIPs in LB (dconde, 17:14:05)
   13. we have two poolls. grp 1 belongs to pool1, grp 2 belongs to pool 2 (dconde, 17:14:37)
   14. it's no longer a chain, it's a GRAPH. (dconde, 17:14:57)
   15. so new semantics. (dconde, 17:15:02)
   16. if we have a WAF betwen pool 2 and grp 2 (dconde, 17:15:41)
   17. if WAF a member of EPG or something else. (dconde, 17:15:51)
   18. as dlenrow said we treat service function as a group (dconde, 17:16:24)
   19. so we are drawing group graph view (dconde, 17:17:11)
   20. do we don't wnt to get stuff confused with application intent. (dconde, 17:17:57)
   21. do we treat them as regular rules to redirect? (dconde, 17:20:22)
   22. probably need to treat tranparent functions differently from non-transparent (dlenrow, 17:21:46)
   23. we want these to be auto provisioned too. (dconde, 17:23:27)
   24. are we a plumber or what? (dconde, 17:23:31)
   25. mickey_spiegel says no everything is transparent. (dconde, 17:34:09)
   26. https://blueprints.launchpad.net/neutron/+spec/adv-services-in-vms (dconde, 17:40:45)
   27. AGREED: we need to bring in others who are more familiar with the services world (Quinn) (dconde, 17:46:38)
   28. we are talking about http://tools.ietf.org/html/draft-quinn-sfc-arch-04 (dconde, 17:48:14)

Action items

1. (none)

People present (lines said)

1. dconde (35)
2. mickey_spiegel (4)
3. odl_meetbot (4)
4. tbachman (3)
5. dlenrow (1)
6. alagalah_ (0)