[ippm] TWAMP YANG data models

[Greg Mirsky <[e-mail]>]

Dear Authors, All
in our meeting in Berlin I've shared my concern with
the draft-ietf-ippm-twamp-yang. These are my comments:

   - Figure 2 presents view of how YANG data model my be used. I believe
   that in SDN environment the Orchestrator may act more like described
   in draft-mirsky-ippm-twamp-light-yang-03 Figure 1 where Session-Sender and
   Session-Reflector are configured directly without Server and Control-Client
   by the single SDN controller/configuration client;
   - even though TWAMP-Control is optional, one possible solution to
   control TWAMP-Test, TWAMP-Test dataa model depends on TWAMP-Control data
   model. For example:

          leaf ctrl-connection-name {
            type string;
            config false;
            description
              "The name of the parent TWAMP-Control connection that
              is responsible for negotiating this TWAMP-Test session.";
          }

in list test-session, which is in session-sender;


   - as I look into session-sender YANG model I don't see it contains
sender-ip,
   sender-udp-port, reflector-ip, reflector-udp-port, and test-packet-dscp.
   I believe that without these session-sender cannot be used to instantiate
   TWAMP test session without the TWAMP-Control;
   - I think that the test session state information provided by
   maintenance-statistics should be extended, e.g. with the container
   current-stats as in draft-mirsky-ippm-twamp-light-yang-03;
   - purpose of parent_connection_XXX in session_reflector is not clear to
   me. According to the last paragraph of section 4.4 four-tuple
{parent-connection-client-ip,
   parent-connection-client-tcp-port, parent-connection-server-ip,
   parent-connection-server-tcp-port} define the test session but that, in my
   view, is only in some scenarios when TWAMP-Test session instantiated via
   TWAMP-Control. I believe that in SDN environment, where TWAMP-Control use
   is not required, identities of Control Client and the Server are not used;
   - grouping maintenance-statistics used in both session-sender and
   session-reflector. If the purpose of the grouping to provide information
   about the measured parameters, then, in my view, it is too limited and
   should be extended. If the grouping is to provide information for
   debugging, then it seems as more proprietary rather than standard data.

Regards, Greg