1. Criteria for Protocol Selection Barbara Stark, AT&T Tim Carey, Alcatel-Lucent Prepared for IETF LMAP WG discussion during 15 December 2014 meeting. 11 December 20141

2. Criteria for Controller Protocol Selection MUST have (protocol will not be considered that does not meet these criteria): There must be a mechanism that allows a Controller to cause a session to be established with a MA. There must be a mechanism that allows a MA to cause a session to be established with a Controller. The protocol session must be capable of being secured using certificates. Page 2

3. Criteria for Controller Protocol Selection Criteria for comparison: How many exchanges are required to send a complete instruction set? (less is better) How many exchanges are required to send a status update? (less is better) Is it possible to provide partial updates? (yes is better) Are there any special mechanisms (other than STUN/TURN/ICE or using port forwarding pinholes, PCP, UPnP IGD, etc.) for NAT/firewall traversal? (asked out of curiosity, but not expecting anything) How many bytes of overhead are required to send a complete instruction set? (less is better) How many bytes of overhead are required to send a status update? (less is better) How widely used is the protocol and/or its protocol elements in mass market devices? (widely is better) What mechanisms exist to ensure interoperability of MA and Controller implementations? (existence of something is better) Are the components of the protocol available as open source? (yes is better) What ecosystem of tools exists for developers implementing the protocol (include tools for data model creation)? (existence of useful tools is better) Is the protocol versionable? (yes is better – or is this mandatory?) – If yes, what is the process for extending the protocol? (for information) What are the encodings supported by the protocol (SOAP, JSON, XML, etc.)? (for information) Page 3

4. Criteria for Report Protocol Selection MUST have (protocol will not be considered that does not meet these criteria): There must be a mechanism that allows a MA to cause a session to be established with a Collector. The protocol session must be capable of being secured using certificates. Criteria for comparison: How many exchanges are required to send a report? (less is better) Does it allow for sending multiple reports in a session? (yes is better) Is there a capability for long-lived sessions. (yes is better) Is compression supported? (yes is better – or is this mandatory?) How many bytes of overhead are required to send a report? (less is better) How widely used is the protocol and/or its protocol elements in mass market devices? (widely is better) What mechanisms exist to ensure interoperability of MA and Collector implementations? (existence of something is better) Are the components of the protocol available as open source? (yes is better) What ecosystem of tools exists for developers implementing the protocol (include tools for data model creation)? (existence of useful tools is better) Is the protocol versionable? (yes is better – or is this mandatory?) – If yes, what is the process for extending the protocol? (for information) What are the encodings supported by the protocol (SOAP, JSON, XML, etc.)? (for information) Page 4