Cell Relay Archive

[James Scott <jscott@net.com>]

hi michael,

martonem@bah.com wrote:

|||> Can UNI 4.0 provide modifications to current SVCs without
|||> breaking the current connection?  That is, the old way of 
|||> changing QoS required a break and then make a new SVC because 
|||> QoS was established solely at the beginning of a connection....
|||> Is this still true

jscott@net.com follows:

||> yes.  in order to accomodate raising a connection's QoS, the
||> connection request would need to be resubmitted to the admission
||> control algorithm at each node in the link.  this implies tearing
||> it down and reestablishing it.

toddmd@aur.alcatel.com follows:

|> UNI 4.0 optinally allows some negotiation during the setup phase
|> of the traffic contract (PCR, SCR, MBS, best effort) by
|> providing an alternative traffic descriptor exclusive-or
|> minimum acceptable traffic descriptor. If in your alternative,
|> you specify best effort, which would be an alternative
|> QoS. The QoS parameters (CDV, maxCTD, CLR, CER, etc.) can be
|> signalled individually, so there's more granularity over
|> UNI3.1.

yes, "during the setup phase," which does not imply dynamic
raising or lowering of QoS.  i explicitly mentioned the case
of "raising" a connection's QoS since this is the more
parthological case for CAC.  any implementation that would
alter a connection's QoS to one with more stringent objectives
without running this request through CAC is either severely
underutilizing its resources or is unstable.

thanks for the additional comments.

cheers,

james