[j-nsp] Question on QoS feature in ERX

[Seán Murphy]

Hi,

I have some questions on the Juniper ERX router and
was hoping someone on this list might be able to help.
The questions pertain to the QoS features of the router.

I've read through the documentation, but I don't quite
understand all of the capabilities of the device.

Here are my questions:

My questions pertain to the qos profile and how it works,
particularly in a context in which there could be many
atm connections on a single interface. I don't quite
understand the group concept and would like to get
my head around this. This is not specific enough, so
here are some more specific questions:

(a) is the group element for scheduling between large numbers
of queues/nodes (i.e. is this what is used for scheduling between,
say, 1000 VCs)? (my guess is yes)
(b) is it only possible to have at most two groups connected to
an output port - the default group and another group? This
grouping is then controlled by the traffic class group, right?
(c) is it possible to implement scheduler mechanisms that
give different scheduler weights to the different groups and
to rate limit both default queues?
(d) is it possible to configure a qos profile for a large number of
VCs such that we can allocate an almost arbitrary scheduling
rate to each individual VC (my guess is no)?
(e) is it possible to segregate the VCs into classes and allocate
an upper rate to each individual VC using some kind of class
profile, so there would be a (scheduler) profile associated
with customers who request up to 2Mb/s, 1Mb/s, 512kb/s
and 256kb/s? (my guess is that this is not exactly possible).
(f) if an atm port is configured such that it uses port mode
queueing does this mean that queueing/scheduling is
essentially performed based on traffic classes? (my guess is
yes)

Any light that could be shed on the above issues would
be greatly appreciated.

TIA,
Seán.