Thread Links Date Links
Thread Prev Thread Next Thread Index Date Prev Date Next Date Index

Re: [RE] Latencies through RE links (was cables)



Pat,

I think hop delay might be the best.

Cheers,
DVJ
David V. James

 

>> -----Original Message-----
>> From: pat_thaler@agilent.com [mailto:pat_thaler@agilent.com]
>> Sent: Wednesday, April 27, 2005 4:51 PM
>> To: dvj@ALUM.MIT.EDU; STDS-802-3-RE@LISTSERV.IEEE.ORG
>> Subject: RE: [RE] Latencies through RE links (was cables)
>> 
>> 
>> The heading of the item was misleading. Buffer and conflict 
>> delays (I'm not sure what you mean by conflict delay as 
>> different from buffer - I assume you are differentiating between 
>> the delay through buffers when there is no other traffic and the 
>> delay when there is other traffic in front of you) are in no way 
>> part of "cable" delay. Cable is just the physical media. Link 
>> delay might include the delay in physical layer devices. I don't 
>> think it includes the delays of items above the link layer such 
>> as switch queues or switch relay delays. Hop delay might be a 
>> title fof something more inclusive. Getting terminology correct 
>> can help in having useful discussion.
>> 
>> Regards,
>> Pat
>> 
>> -----Original Message-----
>> From: owner-stds-802-3-re@IEEE.ORG
>> [mailto:owner-stds-802-3-re@IEEE.ORG]On Behalf Of David V James
>> Sent: Wednesday, 27 April, 2005 10:50 AM
>> To: STDS-802-3-RE@LISTSERV.IEEE.ORG
>> Subject: Re: [RE] Latencies through RE cables
>> 
>> 
>> Geoff,
>> 
>> Thanks for the detailed numbers.
>> I guess it doesn't hurt to use actual numbers, to validate
>> the following statement.
>> >> (This number is much more than the speed of light, ...
>> 
>> I suppose an illustration of conflicting traffic, which
>> can oftentimes be the cause of conflicts, would also
>> be helpful. I believe Pioneer had a slide presentation
>> on this topic, and that had been overlooked in the
>> writeup.
>> 
>> Thanks again,
>> DVJ
>> David V. James
>> 
>>  
>> 
>> >> -----Original Message-----
>> >> From: Geoff Thompson [mailto:gthompso@nortel.com]
>> >> Sent: Tuesday, April 26, 2005 1:45 PM
>> >> To: David V James
>> >> Cc: STDS-802-3-RE@LISTSERV.IEEE.ORG
>> >> Subject: Re: [RE] Latencies through RE cables
>> >> 
>> >> 
>> >> David-
>> >> 
>> >> The cable latency is less than or equal to 570 ns/100 meter link 
>> >> (one way).
>> >> This is a well established number within 802.3 and cabling standards.
>> >> 
>> >> This is based on the 5.7 ns/m figure that is in 10BASE-T 
>> (ref: 14.4.2.4)
>> >> The higher speed links are slightly faster.
>> >> 
>> >> Geoff
>> >> 
>> >> 
>> >> At 12:14 PM 4/26/2005 -0800, David V James wrote:
>> >> >All,
>> >> >
>> >> >Things seem to have been quiet for the last week.
>> >> >Perhaps I could stimulate some discussions on
>> >> >cable latencies?
>> >> >
>> >> >I believe Alexei's presentations have claimed that
>> >> >interactive latencies of 15ms are nearly audible.
>> >> >
>> >> >Since the link delays are only part of the delay
>> >> >equation, this has led some of us to believe that
>> >> >(worst case) per-hop latencies should not exceed
>> >> >0.5ms. (This number is much more than the speed
>> >> >of light, since it includes buffer and conflict
>> >> >delays.)
>> >> >
>> >> >Is there any controversy with using this as a
>> >> >working per-hop maximum delay number?
>> >> >
>> >> >For background material on this topic, please
>> >> >reference pages 15-17 of the following
>> >> >working paper:
>> >> >
>> >> >   http://groups.yahoo.com/group/REInterestGroup/files/
>> >> >   The paper is listed as dvjRePaper2005Apr26.pdf
>> >> >
>> >> >I have spent the last week accumulating content
>> >> >of various slide presentations into the above
>> >> >listed working paper. Hopefully this will be helpful
>> >> >when considering this and other issues.
>> >> >
>> >> >DVJ
>> >> >David V. James
>> >> >
>> >> 
>> >> 
>>