17:04:53 <dconde> #startmeeting ODL-GBP-MODEL 17:04:53 <odl_meetbot> Meeting started Fri May 23 17:04:53 2014 UTC. The chair is dconde. Information about MeetBot at http://ci.openstack.org/meetbot.html. 17:04:53 <odl_meetbot> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 17:04:53 <odl_meetbot> The meeting name has been set to 'odl_gbp_model' 17:05:49 <dconde> Advertisement from Google! You want to buy a new keyboard 17:06:12 <alagalah_> hey 17:06:28 <alagalah_> click click click 17:06:46 <dconde> #chair regxboi 17:06:46 <odl_meetbot> Current chairs: dconde regxboi 17:07:09 <alagalah_> ejkwklj/fsaklaklsaklslkkl;asl;asf;safkl;safkl;sakl;csakl;Tyoping noises 17:07:16 <alagalah_> tap tap tap tap 17:07:22 <tbachman> #info it’s Friday 17:07:42 <dconde> #chair alagalah 17:07:42 <odl_meetbot> Warning: Nick not in channel: alagalah 17:07:42 <odl_meetbot> Current chairs: alagalah dconde regxboi 17:07:43 <dvorkinista> #info Friday morning typing experiment. 17:07:53 <dconde> #chair alagalah_ 17:07:53 <odl_meetbot> Current chairs: alagalah alagalah_ dconde regxboi 17:08:11 <dconde> #topic review model 17:08:19 <s3wong> we have both alagalah and alagalah_ as chairs 17:08:24 <dconde> #info questions on model? by readams 17:08:39 <dconde> #info one thing that may be useful is to look at the latest file 17:08:58 <dconde> #info jan wants deleta 17:09:02 <dconde> #info adding context 17:09:08 <dconde> #info rename role into requirements 17:09:13 <dconde> #info moving stuff around to the back 17:09:17 <dconde> #info change to structure of document 17:09:22 <dconde> #info no semantic change 17:09:46 <dconde> #info change not pushed into repo is the removing of namespace inheritance mapper in the parameters. 17:10:15 <dconde> #info the way inheritance working is that it does not add much. name can be in te matcher itself 17:10:27 <dconde> #info no useful way of using it. 17:11:07 <dconde> #info matcher quality vs. matcher xyz on the same name space was not too useful. It's best to define a different matcher with different semantics. 17:11:19 <dconde> #info you can still over-ride it. 17:11:42 <dconde> #info when you inherit a matcher. there are matcher labels with the addition of NS parameters (NS=name space) 17:11:57 <dconde> #info a lebel in the child EPG inheri fro parent EPG 17:12:12 <dconde> #info if the label with the same name in the child, it replaces it. 17:12:45 <dconde> #info if it is defined in the child, it over-rides the parent matcher. 17:12:57 <dconde> #info can you exclude a label? Yes. 17:13:49 <dconde> #info if you define a label in child of anything, it will override the parent UNLESS you set hte inheritance to exclude, then the label will not be there. 17:13:58 <dconde> #info that is called the inclusion rule. 17:14:39 <dconde> #info dvorkinista says it's like the polymoprhism. 17:15:44 <dconde> #info we can add overrides later. We should not import all semantics of programming languages into this this, just because you can. 17:16:21 <dconde> #info uchau is talking about complex typse. 17:16:37 <dconde> #info yang tools do not support that? can jan confirm that? 17:16:46 <dconde> #info it's just not implemented in ODL? asks uchau 17:17:07 <dconde> #info can we just put chd inside parent in complex types. 17:17:27 <dconde> #info dvokinista says that is discourged in obj oriented systems. 17:17:33 <dconde> #info the model is reasonable, says readams. 17:17:44 <dconde> #info janmedved was there when it was defined. 17:18:10 <dconde> #info this is the best we can do. we can explore extension to YANG. it does complicate things. 17:18:36 <dconde> #info tree structure and the simple way to create it is lending itself to a simple implementation. we are exploring extension to YANG. 17:19:24 <dconde> #info there aer advantages in lookup behavior in doing it that way. 17:20:00 <dconde> #info uchau it is simpler from tree traversal point of view. it would be nice, dvokinista agrees it would be nice. 17:20:30 <dconde> #info we can return link to parent in REST API, for example. 17:20:34 <dconde> #info said readams. 17:20:50 <dconde> #info when you have a ref to someting in a subtree, we can include an actual URI. 17:21:00 <dconde> #info regxboi gets clarified. 17:21:27 <dconde> #info regxboi says RESTconf does not do that today. it's a pointer, not a URI 17:21:45 <dconde> #info we can do an extension , says readams 17:22:12 <dconde> #topic traffic 17:22:19 <dconde> #info traffic within an EPG 17:22:27 <dconde> #info 1) always allowed. 17:22:39 <dconde> #info 2) traffic not allowed, and we need to enable it via a CONTRACT 17:23:02 <dconde> #info dvorkinista says allow is like how VLANs work. 17:23:12 <dconde> #info not allowed is like VDI. 17:23:24 <dconde> #info or do a contract, that satisfied HP requirements. 17:23:57 <dconde> #info challenge - how to define the policy in multi-cast and broadcast. Between EPG we can do unicast 17:24:10 <dconde> #info if we ignore multicast, we can have a peering contract. 17:24:44 <dconde> #info define a set of sibjects within EPG 17:24:49 <dconde> #info or define a peer. 17:25:34 <dconde> #info we need a way to apply a contract by stating it's a peer -- no direction. 17:26:06 <dconde> #info we can say something like EPG -- all elements are allowed, or we can say all classifiers are interpreted as bidi as a peer. 17:26:18 <dconde> #info since we do not want to define a new type of classifier with no direction. 17:26:55 <dconde> #info we can prefabricate things, and put tha tinto the group itself. 17:27:06 <dconde> #info but that complicates things, says dvorkinista 17:27:15 <dconde> #info we need a differet model 17:27:24 <dconde> #info we need a new MODE. 17:27:43 * tbachman thinks regXboi might be raising his hand 17:28:16 <dconde> #info regxboi says - his gut reaction is lot of complexity for small incr. gain. 17:28:29 <dconde> #info readams -- outside a Datacenter, we need it. 17:28:34 <dconde> #info regxboi cries fowl 17:28:53 <tbachman> foul even :) 17:29:12 <dconde> #info dvorkinista says we have requirements for different use cases from yesterday. 17:29:15 <dconde> cluck 17:29:18 <tbachman> lol 17:29:57 <dconde> #info less moving parts is better? 17:30:15 <dconde> #info regxboi says -- to degenerate conclusion, we need to to talk about contracts on EP? 17:30:23 <dconde> #info people say no, it's a group attribute. 17:30:45 <dconde> #info you either turn all OFF or ON, or provide a contract. 17:31:10 <dconde> #info we can have a bunch of app servers, and allow no HTTP or SSH, and only allow control protocols. 17:31:40 <dconde> #info we need a model on the EPG for allow or deny. 17:32:18 <dconde> #info dvorkista this enables unified communications - sessions between different callers (End Pts) 17:32:49 <dconde> #info boundaries are a TENANT. 17:33:07 <dconde> #info cross tenant mutations -- have particular mechanisms for that. 17:33:27 <dconde> #info uchau needs clarification. 17:33:54 <dconde> #info EP are producer/consumer. we are introducing a ShORTcut since it's highly inconvenient. 17:34:00 <dconde> #info we want to emulate VLAN behavior. 17:34:15 <dconde> #info ans also the VDI behavior when no endpoint can talk to each other. 17:34:34 <dconde> #info traffic belongs are defined as something within a group. 17:34:50 <dconde> #info that makes it easier to understand. 17:34:58 <dconde> #info we are folding contract concept within an EPG? 17:35:13 <dconde> #info you will need to select contract by name 17:35:30 <dconde> #info so we cannot enable anyone else to name and consume traffic. 17:36:06 <dconde> #info are we stepping into restrictions? 17:36:22 <dconde> #info it's not an issue of who gets to consume or not. 17:36:32 <dconde> #info semantics for peer A can talk to Peer B 17:36:47 <dconde> #info where does that get folded? in a contract. 17:36:59 <dconde> #info uchau says do we have a special selector 17:37:17 <dconde> #info we will have a special peer thing. for unified comm. we want to specify it once. 17:38:03 <dconde> #info a select pts to a contract and then within a contract . how do we do the peer-to-peer association 17:38:08 <dconde> #info a group specified a session. 17:38:23 <dconde> #info if there are two EP, then there are 2 EP in a group. 17:39:28 <dconde> #info we define a contract for the ENTIRE group. 17:39:38 <dconde> #info we are not talking about End point to End point contracts. 17:40:03 <dconde> #info uchau needs clarification 17:40:21 <dconde> #info subgorup of two users? Not really. 17:40:47 <dconde> #info lets say. session represents a group 17:40:56 <dconde> #info a set of ports that members of the group will be talking to. 17:41:24 <dconde> #info they will use the contracts that ….specify 17:41:35 <dconde> #info we are switching to the whiteboard. 17:42:40 <dconde> #info INTRA group comm via a PORT is a simple use-case 17:43:08 <dconde> #info now a diff example 17:43:18 <dconde> #info a DB cluster 17:43:29 <dconde> #info a contract called SQL is applied 17:43:46 <dconde> #info we want to restrict - to clustering protocol only. 17:44:04 <dconde> #info we say there is a peer within group pointing to clustering protocol contract. 17:44:24 <dconde> #info it is similar to contracts. it is scoped to group only 17:44:56 <dconde> #info what about direction? 17:45:55 <dconde> #info if a contracts has requirements/capability matchers. how do we eval them? 17:49:13 <dconde> her regxboi can you scribe? 17:49:21 <dconde> s/her/hey/ 17:50:15 <dconde> #info any reason for a peer target selector? no 17:51:20 <dconde> #agreed name selectors are sufficient. 17:51:37 <dconde> #info how do we have tie breaker rules? 17:52:07 <dconde> #info if there are two, then we apply them in order. 17:52:17 <dconde> #info multiple contracts in scope is OK. 17:53:24 <dconde> #info under rule application, go read that. let readams if that's wrong. 17:53:39 <dconde> #agreed we need to go read it and then talk next week 17:54:06 <dconde> #info subject have order. 17:54:19 <dconde> #info we have ordering rules for how they are applied 17:54:31 <dconde> #info no mixing of rules in ACLs. that will not work. 17:54:43 <dconde> #info read rule applicaton secton and subjects under inheritance. 17:55:17 <dconde> #info regxboi says we are running out of time. he cannot make it to that one. 17:55:28 <dconde> #info regxboi delegates to mspiegel. 17:56:05 <dconde> #help 17:56:38 <dconde> #info uchau wants policu def in JSON fmt. 17:56:55 <dconde> #info please read Wiki and bring it up under RESTCONF now. 17:57:22 <dconde> #info ask readams postings under mailing list. 17:59:29 <dconde> #info if you go into swagger api docs, if you look at what it says, then only a small subset will work. so jan will look 17:59:47 <dconde> #info put and post have different impl, and we think semantics are slightly different. 17:59:57 <dconde> #info not high priorities. 18:00:38 <dconde> #info go look in YANG model for the fields in a JSON obj under relevant type 18:01:07 <dconde> #info use FireFox not Chrome 18:02:35 <dconde> #action readams will send link to mailing list on the use of REST API. -- it's better than curl from cmd line. 18:03:13 <dconde> #endmeeting