Date: Thu, 13 Jun 96 10:21:22 EDT From: Anil Rijsinghani 13-Jun-1996 1013 To: jlarson@fnal.gov, p8021@hepnrc.hep.net Cc: anil@netcad.ENET.dec.com Subject: 802.1p/D3 Ballot SUBJECT: P802.1p/D3 - Traffic Class and Dynamic Multicast Filtering Services in Bridged LANs __X___ I disapprove for the following reasons. Anil Rijsinghani ---------------------------------------------------------------------- 1. Unicast address registration should be removed. Also suggest explaining in replacement text why this was done (perhaps by borrowing from some of the Wakefield presentations). 2. The "recommended" use of two priorities is unusual given that the spec allows for more. Suggest changing this to "minimum" (at the risk of stating the obvious), or changing the max number of priorities to 2. 3. Recommendations for some parameters may usefully be less than the max: for example, when the range of a hello timer is from 1 second to 10 seconds, and the recommended value is 2 seconds, that is understandable; however the recommendation for 2 priorities confusing -- it is similar to recommending a lower switching rate pps than the maximum possible for a switch. While in practice it may not be possible, for economic reasons, for a low-end switch to actually implement the maximum, a standard should not actually recommend lower performance. 4. Priority for multicast traffic with 802.1p needs to be carefully re-evaluated. Given the change of mindset that started with the Wakefield meeting to incorporate QoS in VLANs, we need to make sure that we are not writing two standards at the same time that allow you to do the same thing. At best this will be confusing, at worst there will be interoperability pitfalls. Further consideration of this issue is called for. 5. A pure priority-based scheme is not practically useful in some scenarios. More consideration needs to be given to the issue of bandwidth reservation at different priority levels. At the least, the draft should mention that this issue has been discussed and is expected to be pursued at a later time.