SG15-LS1: LS/o/r on ONF OIMT early draft TR-512.A.15 on Controller Lifecycle and Security (reply to ONF-LS010 – TD801/WP3)
- LS was received on 22 March
- The attachment is a marked-up copy of draft TR-512.A.15.
- Nigel Davis To post the LS on the liaison wiki page
Stephen walked through the Q12 & Q14/15 feedback on draft TR-512.A.15
- The TR-512.A.15 draft was considered in the Q12 & Q14/15 February E-meeting
- 3.2 The figure: Should reference Fig. 6-1 of ITU-T G.7701 Common control aspects (2022)
- 3.2 Last figure: The assumption of entity relationship in the server to client contexts needs to be clarified.
- 3.3 The Controller overview figure: Suggest re-labeing for consistency. Suggest to consider additional content of RDB.
- 3.4.1 Client: Suggest to clarify that a client context is not aware if its cleint uses the CM for a network slice.
- 3.4.2: "link" should be "interface"
- 4.1 Platform creation: Might want to breakout CPU, memory, storage explicitly from compute resources. Still under the heading of compute resources.
- 4.2 Adjacent controllers: Need clarify what adjcent mean, client/server or peer?
- 4.3 Need more discussion on Local planned resources and Local transport resources; In what contect does the Links belong to?
- 6.2 Multiple views: Clarify CPI. Does CPI contain one or multiple sessions
- 6.2.1 View translation: Commonality with descriptions in G.7716.
- 7 Controlled access session references: Similar to call control in G.7701. There are existing protocols for peer authentication and closed user groups etc.
- Larger topic - Compute.
- The draft provoked extensive discussion in Q12 & Q14 on the computing resources, the metrics that could be used to describe compute resources, and how they might be allocated.
- What is the quantity of compute resources; how to allocate that
- MB: This first draft is try to identify the areas that we need to address.
- MB: How to prevent a client from accidentally overload the compute resource of the controller impacting the performance. Then what do we try to constraint - memory, storage, etc.
- SS: 6.2.1 compute resource recursive allocation: following networking recursion could be helpful, but for compute resources it will very limited in depth.
- SS: The boundary between compute & networking is becoming blurry.
Follow up:
- More dialog between ONF and Q12/Q1, specially on compute.
- Q12/Q14 e-meeting on April 12 & April 14
- ONF should respond, even partial result
- Need at least one more iteration of the draft