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

Re: SONET/Ethernet clock tolerance


A major difference between T1X1.5 and 802.3ae is that T1X1.5 is encoding
over SONET, supporting all SONET clock, jitter, and management
requirements, while 10 GigE WAN-PHY uses Ethernet compatible timing. This
makes these two standards very different. It also address the issue of
having a simple low cost WAN access media (10 GigE) and at the same time
having a full fledged SONET compliant version.



At 02:24 PM 3/31/00 +0900, Osamu ISHIDA wrote:
>I think this is one of the biggest inter-culture gap between LAN and 
>traditional (or Telecom) WAN.
>At 8:35 PM -0600 00.3.30, Roy Bynum wrote:
>> I am not sure why you are concerned about the regenerators budget timing
>> specifications.  Loop timed or derived timed systems run on the clock
>> specification of the incoming clock.
>In datacom regenerators or media converters, they can easily re-clock 
>with its local clock source by simply inserting/removing IDLE character.  
>Hence they worry about the jitter specification only between each two 
>regenerators.  This makes their PLL design very easily and inexpensive.
>In Telecom world, on the other hand, we must be SYNCHRONOUS once we 
>have packed the SONET frame.  This means SONET regenerators have to 
>re-use the upstream incoming clock for the downstream signal to 
>transmit.  This makes our PLL design very difficult, since we have to 
>worry about jitter transfer at each regenerator and about its 
>accumulation all over the regenerator chain.  I know that many 
>traditional SONET regenerators even don't use PLL to meet this stringent 
>jitter transfer specifications; they use more expensive SAW filter to 
>extract the incoming clock.  Please consult with ITU-T Recommendation 
>G.958 Clause 9.3: Jitter performance.
>Once again, I agree that SONET-compliant PHY makes sense since it can be 
>connected directly to install-base SONET regenerators.  Here I support 
>Nortel's T1X1.5 proposal (Ether on SONET) since I have no other choice 
>at present. (Or will anyone propose 64/66 on full-SONET?)
>I don't believe SONET-framing make sense just for the OAM&P signaling 
>to Ethernet Line Terminating Equipment (ELTE).  SONET framing is 
>optimized for SYNCHRONOUS transport systems, and hence we should pay 
>too heavy tax (read it pointer manipulation, SONET re-framing, or extra
>buffers) when we want to change our clock for relaxing the jitter budget.
>This is the background of my 10GENIE proposal (Layer-1 signaling in IPG) 
>at Albuquerque.  I hope to see brand-new WAN for datacom, since I believe 
>this is the most economical way to utilize our fiber networks.
>Best Regards,
>Osamu Ishida
>NTT Network Innovation Laboratories
>TEL +81-468-59-3263  FAX +81-468-55-1282
Paul A. Bottorff, Director Switching Architecture
Enterprise Solutions Technology Center
Nortel Networks, Inc.
4401 Great America Parkway
Santa Clara, CA 95052-8185
Tel: 408 495 3365 Fax: 408 495 1299 ESN: 265 3365
email: pbottorf@xxxxxxxxxxxxxxxxxx