| Thread Links | Date Links | ||||
|---|---|---|---|---|---|
| Thread Prev | Thread Next | Thread Index | Date Prev | Date Next | Date Index |
|
Regarding my comment below on the one oMACEntity to many oPHYEntity objects, the odd thing about the existing diagram is that the relationship of oPHYEntity to oAutoNegotiation is one to one (through oMAU) and
it seems like it should be many to one – all the PHYs that share an autoneg having a containment relationship between their oPHYEntity objects and the oAutonegotiation object. Here is the text below on oPHYEntity that explains that there can be many PHYs but only one active PHY at a time: oPHYEntityIf oOMPEmulation is implemented, oPHYEntity is contained within
oOMPEmulation. Otherwise oPHYEntity is contained within oMACEntity. Many instances of oPHYEntity may coexist within one instance of oMACEntity; however, only one PHY may be active for data transfer to and from the MAC at
any one time. oPHYEntity is the managed object that contains the MAU, PAF, and PSE managed objects in a DTE. I don’t see anything that explains why the containment from those multiple PHYs is one to one to stuff below them that there is only one of is one to one. Maybe Howard can explain it. Pat From: Pat Thaler [mailto:pthaler@xxxxxxxxxxxx]
Hi Marek, Please look at the existing 30-3 drawing. The one to many relationship of MAC to PHY comes from the existing relationships. My recollection is that, for management purposes, multi-speed phys are modeled as multiple PHYs attached to the
same MAC only one of which can be enabled at a time (for transmit and active operation – during auto-neg multiple receivers can be active for parallel detect).
If you don’t think that is right, it would be a maintenance issue – not 802.3BR. I agree with the rest of your comments except that I’m leaning toward the second version of the drawing where the second oMACEntity is contained in the other oMACEntity and we don’t need to show a 2 to one relationship. It also has the
advantage of not needing any extra elements beyond the containment relationship to show that the two MACs are a pair. Regards, Pat From: Marek Hajduczenia [mailto:marek.hajduczenia@xxxxxxxxx]
Hi Pat, I have one comment on the proposed drawing: it seems that a single oMACMergeEntity maps into two oPHYEntity instances – as far as I understand, we have two MAC with MAC merge layer, but not two PHYs. I think
it should be a 1:1 relationship between oMACMergeEntity and oPHYEntity. Some more answers inline Regards From: Pat Thaler [mailto:pthaler@xxxxxxxxxxxx]
Here is my initial try on a relationship diagram for MAC Merge. I haven’t done one of these before.
For now this is a diagram that applies only when MAC Merge is present. If adding MAC Merge to Figure 30-3 is done similar to this diagram, then I think this could be merged into Figure 30-3 in a way similar to what was done with MAC Control.
I.e. Add a one to many relationship between oMACEntity back and oPHYEntity in and in the description of oPHYEntity state that it is contained in oMACMerge when oMACMerge is present and otherwise it is contained in oMACEntity.
I also considered having oPHYEntity always contained by oMACEntity as it currently is, but then which MACEntity of a MACMerge sublayer own it. We don’t have any many to many or two to many relationships defined. Also, this is more similar
to the way MAC Control was added. I have a number of questions about this drawing: There are a number of questions about this diagram: Should the deprecated objects be removed since it is a new diagram? [mh0814] I suggest to submit a comment against 802.3REV and remove deprecated elements – that is not within the charter of IET to fix that.
Should a new symbol be added to indicate a two to one relationship instead of using the many to one relationship?
[mh0814] It would not hurt to extend the current drawing model, where we add a number next to the arrow to indicate the number of instances: (2), (1), or (n) Should this relationship be shown in some other way?] [mh0814] I believe it is correct as it is shown right now.
Should oResourceTypeID stay where it is or should it be contained by oMACMergeEntity? If it stays where it is, there may be one for the eMAC and one for the pMAC. Note that oResourceTypeID is specified in 802.1F, a withdrawn standard.
Because of that, I lean toward leaving it where it is. [mh0814] again – not a problem we need to fix under IET …
|