The IP over ATM Mailing List Archive by date

[hiroshi@ctr.columbia.edu (Hiroshi Esaki)]

    mwg> The ATM folks think it's so simple that it
    mwg> will have no market, while many IP folks want nothing more complex
    mwg> from ATM than FIFO queueing with big buffers (which would be a
    mwg> satisfactory implementation of UBR.)

   jh> i have nothing against simplicity, but you make implementing ubr too
   jh> simple.  even in case of ubr, one would like to have at least some
   jh> degree of fairness and to avoid dropping of cells, unless you at the
   jh> same time drop all cells of the same frame.  large buffers alone don't
   jh> provide this.

FIFO with big buffer will be faur for every body, I think. 
Also, the significant cell loss for UBR is kind of mis-enginnering. 
This is competely the same in the Ethernet. 

Hiroshi Esaki