#opendaylight-group-policy: ODL-GBP-MODEL
Meeting started by dconde at 17:10:56 UTC
(full logs).
Meeting summary
-
- readams multiple contexts per broadcast
domain (dconde,
17:11:38)
- every domain has a subset associated with
it? (dconde,
17:13:29)
- mickey asking whether we are separating various
constructs (dconde,
17:19:09)
- FD vs BD -- flooding domain or a forwarding
domain? (dconde,
17:20:10)
- we were gettng confused (dconde,
17:20:18)
- BD = Bridge FD = Flood domain (dconde,
17:20:46)
- multiple FD per BD is OK (dconde,
17:21:19)
- we were confused with terms. now it's all
clearer (dconde,
17:22:27)
- subnets can be part of these constructs.
(dconde,
17:22:38)
- subnet masks to BD, FD in 1:1 (dconde,
17:22:54)
- can we make it orthogonal? (dconde,
17:23:13)
- we can make it with NO L2 domain (dconde,
17:23:35)
- all L3 is possible. (dconde,
17:23:53)
- if we have bridge but not flood domain, then
what? (dconde,
17:24:45)
- is it 1:1? (dconde,
17:24:50)
- then we don't want to require specifying
both. (dconde,
17:25:02)
- two cases. Only L3 is one case. no BD
(dconde,
17:25:47)
- just like Amazon (dconde,
17:26:05)
- then do we have two case? Multiple
subnets? (dconde,
17:26:22)
- EPG belongs to a subnet. (dconde,
17:26:31)
- L3 context gives a namespace. (dconde,
17:26:57)
- with no gateway, you cannot talk to outside
without NAT'ing. (dconde,
17:27:45)
- like a classic overlay case. (dconde,
17:27:51)
- if you have two name spaces? (dconde,
17:29:00)
- what is a L3 domain -- then it's just a set of
IP addresses? (dconde,
17:31:07)
- some customers can achieve security without
creating a sandbox. (dconde,
17:32:36)
- what do we want -- context vs. isolation.
these are independent attirbutes (dconde,
17:34:04)
- everyone who has done this sort of thing has
defined a similar construct (dconde,
17:34:29)
- L3 context has several Bridge domains, and we
assume there is BD, do we want an automatic FD. (dconde,
17:35:31)
- we talked about whether rendering can do "best
job" and still fulfill, vs. unable to do (dconde,
17:51:52)
- like running compiler and looking at machine
code (it's idiosyncratic output, but still will work) (dconde,
17:52:36)
- priorities of static vs dynamic contract -
sanjay -- he will upload PDF of the PPT (dconde,
17:53:47)
Meeting ended at 17:55:45 UTC
(full logs).
Action items
- (none)
People present (lines said)
- dconde (35)
- odl_meetbot (4)
- tbachman (2)
- regXboi (1)
- alagalah (0)
Generated by MeetBot 0.1.4.