Cell Relay Archive

[danw@bas.army.milNOJUNK (Daniel J. Wentzel)]

Hello all,

     I've been asked to teach a course on the basics of ATM, but have
only had a single class on the subject myself (plus reading many books
and this fine newsgroup :) ).  In preparing my lesson plans and
experimenting with some of our switches, I've come across some questions
of my own (and I don't want to look too clueless to my students:))  See
what you think and ANY help is appreciated:

1.) The current use of the GFC field is supposed to be undefined from
what I've read.  Is this still true and are there any proprietary flow
control schemes implemented at a UNI using the GFC nibble?

2.) Does a VPI = 0 automatically indicate VP switching?  Or is this
implementation specific?

3.) I understand that F4 flows are used for VP termination management
(management on a particular VP of some user VPI=x, VCI=3 or 4).  VCI=3
indicates segment flow and VCI=4 indiates end-to-end flow.  However, if
VPI values only have LOCAL significance, how can there be an end-to-end
flow to manage (VCI=4)?

4.) Does UPC perform any tagging of F4 or F5 flows?  Are these flows
always queued first or are they subject to queuing delays to meet
contract obligations?

5.) If there are 3 different formats for the ATM End Station Address and
different organizations can assign values to the beginning 13-octets of
the address, doesn't that limit the probability that the full address
will be unique?  The onus for uniqueness really seems to fall to the ESI
section of these address formats.  Since this is 6 octets and since MAC
addresses for IP devices are 6 octets won't ATM face the potential of
duplicate addresses just like (supposedly) MAC addresses?  Couldn't ATM
face this same 'shortage'?  This seems like a problem for global
uniqueness unless there is a records body tracking this information....

6.) During call setup, how does the egress switch (the switch on the far
end of the network where the called party is locatd) indicate to the
called party which VPI/VCI it should use?  There seems to be an
information element within the setup frame for this purpose, but that
would mean the egress switch would have to 'intercept' the setup message
and pass it off to it's control processor to determine an appropriate
VPI/VCI.  Is this how setup is done?

7.) Is it possible to build a bi-directional call using a single setup
message from calling party to called party or does the called party have
to repeat the setup action in the other direction for this to be done?
If the latter, how does the calling party indicate to the called party
to initiate this action?

TIA

DJW

REMOVE NOJUNK from the end of the e-mail address for proper delivery.