#opendaylight-group-policy: UseCases

Meeting summary

1. 1. https://meetings.webex.com/collabs/#/meetings/detail?uuid=M4NPPO6UULL1B0DFPK2HQ9VZX8-9VIB&rnd=563175.79385 (dconde, 20:06:05)
   
   
2. dvorkin sent exceptions list (alagalah, 20:06:46)
   1. Cisco user vs Sales users need different access (alagalah, 20:09:31)
   2. sanjay says that some users need lower security and some need hi security (alagalah, 20:10:32)
   3. sanjay talks about a hierarchy, dvorkin say lets make subjects hierarchical, no contracts (alagalah, 20:12:42)
   4. examples are getting more complex, dvorkin says classifiers into clauses (alagalah, 20:15:10)
   5. sanjay says single list of rules, but dvorkin says this is what we are trying to avoid..... but it complicates more mainstream cases (alagalah, 20:16:30)
   6. readams proposes that we have a list of actions (alagalah, 20:19:56)
   7. readams wants to be careful of defining an exponential # of rules. (alagalah, 20:25:29)
   8. dvorkin says we can make an exception at the clause level (alagalah, 20:28:58)
   9. https://wiki.opendaylight.org/view/Group_Policy:Sub-Groups:REQUIREMENTS#Documents (tbachman, 20:32:27)
   10. dvorkinista is going to modify the slides (alagalah, 20:32:32)
   11. dvorkinista is changing the PPT (alagalah, 20:34:40)
   12. dvorkinista is going through this email, changing topic (alagalah, 20:50:01)
   
   
3. Dvorkin's over-ride model (alagalah, 20:50:19)
   1. new model that defines clobber etc for contracts (alagalah, 20:51:11)
   2. lenrow points out that its more of a stack type structure where you pop stuff off (alagalah, 20:52:18)
   3. lenrow points out that AAA system has control over where things get added or clobbered (alagalah, 20:54:01)
   4. mickey_spiegel asks question about clobbering QOS (alagalah, 20:56:23)
   5. dvorkinista goes to get note book with thoughts documented (alagalah, 21:02:34)
   6. dvorkinista says that exception rules are for planning for the unplannned.... but the model should handle normal operations (alagalah, 21:04:24)
   7. dvorkinista says that normal intent should be maintained until exceptions occur with overrides (alagalah, 21:05:21)
   8. lenrow is concerned about the stack of overrides and how they resolve (alagalah, 21:05:49)
   9. I think of this as a "parent-override" the tenant is being overruled by an ancestor, no? (lenrow, 21:14:12)
   10. unplanned policies can come in any form, says sanjay (dconde, 21:31:09)
   11. dvorkinista infrastrcture rules that has nothing to do with user intent can happen --like DDoS detection or firewall rule (dconde, 21:31:46)
   12. for data center scenarios , this should be sufficient says dvorkinista (dconde, 21:33:02)
   13. lenrow says it's out of scope of GBP (i.e. for simple use case) (dconde, 21:33:37)
   14. sanjay will work based on it. (dconde, 21:33:51)
   15. ball passed back to lenrow (dconde, 21:34:07)
   16. lenrow seeking volunteers to write up ideas and prsent in future meeting. UC use-cases has gone as far as we can without a straw man (dconde, 21:34:54)
   17. such as on-demand traffic engineering use cases. (dconde, 21:35:26)
   18. we need a volunteer for a one page description (dconde, 21:35:53)
   19. we need more interction with SFC in ODL and SFC in IETF and invite them to these meetings (dconde, 21:36:15)
   20. and write a paper on flow rules are written in each format. (dconde, 21:36:41)
   21. dvorkinista has talk w/ Paul Quinn. (dconde, 21:36:53)
   22. there are OK with GBP specifying high level intent. (dconde, 21:37:20)
   23. lenrow asks who is north and south? (dconde, 21:37:32)
   24. dvorkinista says GBP is north of them (SFC) (dconde, 21:37:41)
   25. lenrow what about competing rule writers? (dconde, 21:38:21)
   26. dvorkinista says hes, but wants to defer that. (dconde, 21:38:35)
   27. an awful lot of meeting. let's see in next few days (dconde, 21:40:03)

Action items

1. (none)

People present (lines said)

1. alagalah (27)
2. dconde (23)
3. odl_meetbot (7)
4. Bhushan_ (3)
5. tbachman (2)
6. lenrow (1)
7. s3wong (1)
8. busha (0)
9. mickey_spiegel (0)