I have attached an amended version of the QoS table based on the small amount of input received to date. There are several changes:
1. Some columns have been deleted: there were no entries for Jitter, Wander, and Other. In general jitter can be handled by a de-jitter buffer which is simply folded into the total delay over the link. Wander, or long-term jitter, is not really an issue with modern clocking techniques, and nothing in the Other category showed up.
2. I have changed the lower quality voice entry to range from 6 kbps (as in G.723.1) to 16 kbps, since the higher rates such as 32-kbps ADPCM and above are covered under high quality voice. Also, I've added ranges for ATM going up to 155 Mbps.
3. Per suggestion, some columns have been deleted. IP is now divided into best effort, time-critical, and non-time-critical services. I also changed ATM to match in a rough way, with classes for constant bit rate, variable real time, and other. From our perspective, there isn't any useful difference between ABR and variable non-real-time, since transmission in a multipoint system must be controlled in any case. For the same reason, UBR doesn't apply since no one can transmit in an uncontrolled way.
Please send feedback on these changes.
Also, the door is now open for items going beyond the scope of this table: text explaining the table items, and appendices to cover items such as application classes.. Application requirements are end-to-end, in distinction to our QoS table which covers only the budget for the wireless access link
Finally, it appears that Scott Adams may be on to us. See Dilbert at
The deadline for getting everything done, including the issues delegated to the QoS group at the August 4-6 meeting, is September 24. I'll forward a list of those issues shortly.http://www.unitedmedia.com/comics/dilbert/archive/cal-22.html
Have a great Labor Day weekend, but send your feedback first!
Regards to all,
Jim