Cell Relay Archive

[murray@pa.dec.com (Hal Murray)]

In article <6e5jee$200@alca.helsinki.fi>, vhalkka@cc.helsinki.fi writes:

> Do you mean, that when the DAGGL-BA card receives unicast
> cells it knows where to send them through the crossbar,
> but when the cells come from a BUS VC the cells are sent
> to the master line card which then makes a number of copies 
> and sends them to every receiving port, one by one ?
> 
> Or is the receiving slave DAGGL-BA able to this ?

I don't know.  I mostly worked on the prototype hardware
rather than the software/firmware.  The hardware could do it
either way - the cost is the same.

I'd expect the receiving card to do it. That would reduce
the workload on the master card.  [But what I expect doesn't
always happen.]

You could find out by running some experiments.  Setup a
UBR multicast VC through a line card and run a test
over it to measure the bandwidth.  Then setup and run
a similar test over another line card.  If the master is
doing all the work the throughput on the first test will go
down when you start up the second test.

-- 
These are my opinions, not necessarily my employers.