Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index



Hi Ryan,

I disagree with some of your statements.

>  By creating fixed timeslots in the upstream you are changing the nature of Ethernet.   
>  Now the maximum bit rate of one station to burst upstream is limited to its timeslot.   
> This creates asymmetry which has never been an explicit form of Ethernet.

Ethernet is already defined for both full-duplex and shared media.  In any shared media network, a node will be getting a share of bandwidth (a portion of the bandwidth).  Using a time slot algorithm is one way to share, CSMA/CD is another.  Both are mechanisms to *share* the media.  So it is not correct to suggest that limiting bandwidth below full rate has never been an explicit form of Ethernet.   Ethernet is very comfortable with shared networks historically.   A new algorithm to do this over 10 km of optical P2MP is what is new.   (By the way, are you opposed to asymmetry in EoVDSL?)

 > I believe according to the All
optic presentation this would be 25 - 50 Mbps/ station.

It depends on the timeslot size, and number of slots assigned to each ONU, and number of ONUs on the PON.  For a 1000 Mbps 1:16 PON, bandwidth per ONU is ~60 Mbps.  At the extreme, one ONU could have all time slots and all the bandwidth.   The Alloptic presentation was simplified, on purpose, to show that one can do EPON in the PHY layer without complexity.   Timeslots, and the ability to do dynamic bandwidth distribution with the timeslots can be discussed.

>  If you increase the latencies across the media ten fold, is it still Ethernet?   

Does Ethernet define latency - I missed that in the spec.   For EPON, there is no latency added in the downstream.  Latency can be kept to a few milliseconds in the upstream for time critical traffic.  Can you explain why this is a problem?

>  The end user will perceive a difference in service. 

Maybe we should build ATM PONs with QoS .....  just kidding.