| | | | - UML of "IEEE 802.1Qcx CFM YANG"
- Latest version: oimt2018.BZ.001.06_PapyrusIeeeCfmOxygenWorkspace_180731.zip
- Changes from the 2018.07.03 version
- Applied the latest {xor} and Bits mapping rules
- The mapping of Bits is still under discussion in ONF, where two options are considered.
- Four UML models, one per YANG module
- New classes: MaintenanceAssociationGroup, CfmOperation, LoopbackReply, MaintenanceAssociationMepList
- Action replace RPC
- New Datatype: ManagementAddressGrouping
- UML diagrams
- Kam recap the comments, on the CFM YANG & UML, from the Q14/15 IM/DM coordination meeting (8/27) discussion. Scott has provided these comments to IEEE through its letter ballot comment process.
- Why "Grouping" at the end of the label ManagementAddressGrouping. the usage is not consistent across all the datatypes
- What is the intent of the paths from MaintenanceAssociationGroup to MaintenanceDomain and MaintenanceAssociation?
- To identify a unique <MD,MA> pair.
- Why between Cfm and the 0..* MD, there is the middle level of container MDs (which has only a description). On the other hand, between the MD and the 0..*MA, there is no such middle level of container MAs.
- In the association between MaintenanceAssociationGroup and MaintenanceAssocation, the cardinality of _maName end should be 1..* (i.e., in the Yang file, line 854, leaf ma-name should be leaf-list, instead of leaf), otherwise it supports point-to-point maintenance association only. If IEEE really supports only point-to-point, then it doesn’t need to have this MaintenanceAssociationGroup) class. Note that ITU-T needs the support of point-to-multipoint maintenance association (e.g., rooted multipoint).
- Comment on the re-engineering from the Q14/15 IM/DM coordination meeting (8/27) discussion
- The association ends from CfmOperation to MaintenanceAssociationGroup and Mep should be mandatory and marked as PartOfObjectKey. Bernd will update the UML
- Questions raised:
- Does MaintenanceAssociation represent MEG or ME?
- Is MEG represented by MaintenanceAssociation or MaintenanceAssociationGroup?
- The Yang statement "list maintenance-association-mep-list" is currently mapped to the MainenanceAssociationMepList object class. But semantic wise, "list maintenance-association-mep-list" is just a list of Mep IDs. The Yang list statement is unnecessary complicate
- MAMepList 1 – 1 MepDb * – 1 Mep. Shoudn't it be MAMepList 1 – * MepDb * – 1 Mep instead?
- Consideration:
- Can/should we improve/simplify the mapped UML?
- The current re-engineered UML is a “literal” translation of the YANG. Therefore the UML model has many artefacts that are unnecessary from UML modelling point of view
- For examples:
- Cfm/MDs/MD (i.e., Cfm contains MaintenanceDomains, which contains MaintenanceDomain). Here the extra level containment is unnecessary.
- MaintenanceAssociationMepList is unnecessary if MaintenanceAssociation directly contains Mep
- As an exercise, we could improve the re-engineered UML, which will then be translated into YANG and compared with the original YANG to see if any information is lost.
- Noted that at the end of the day, Mep and Mip are the touch point classes for augmenting the G.8013/Y.1731 Carrier Grade Ethernet OAM function.
|
