Cell Relay Archive

["Johannes Krohn" <KROHN@ffm.pandacom.de>]

If you think of "packet over sonet: ringing up Speed", Data
Communications 3/98 by Victor R. Parente
(http://www.data.com/tutorial/sonet.html) , then that definitely is
no good point to start reading. The author makes a good point in
describing fast protection switching as a advantage of SONET but on
the other hand does not know how to do his math when comparing
effective bitrates of IP over SONET and IP over ATM over SONET.

The author does not take into account that even when using packet 
over SONET as described in RFC 1619 the octet stream is maped in the 
SONET SPE which means that the effective bandwidth on OC3 is reduced 
by the factor of 260/270 (section and path overhead in SONET/SDH 
framing) resulting in a effective bandwidth of 149,76 Mbps for SONET 
over PPP in a STS-3c framing. Further on the author forgets to take 
into account the overhead introduced by using the HDLC framing 
(no address and control field compression defined in RFC 1619) while 
on the other hand calculating the factor of 48/53 for the ATM 
payload. It@s not that he is completly wrong in pointing out that 
packet over SONET introduces less overhead than IP over ATM over 
SONET, but by stating completely wrong numbers he disqualifies 
himself in a way and introduces the impression of him beeing very 
biased. 

By the way, the author in figure 2 marks some links as beeing "OC3c". 
This in my opinion is a typical error made in using the right names 
for the right things. As stated in the specs OCn simply is the name 
for a optical carrier level where STS-n is the correct naming for the 
used framing. So the correct name in the figure would have been 
either STS-3c (meaning concatenated SPE) or OC3.

The author then points out that by using protection switching on a 
full OC12 SONET ring all links are fully protected against circuit 
failure. This is not only true for a SONET only network but can also 
be used in combination with ATM by building a OC12 SONET ring with 
STS-3 ADMs and then connecting ATM switches to the STS-3 tributary 
links. Then the protection switching mechanism of SONET protects the 
STS-3 tribs on the PHY layer transparently to the ATM equipment. To 
be fair it should be stated that fast protection switching needs a 
priori planning of protection trails through the SONET resulting in 
need for double of the protected tribs bandwidth in the net. Or the 
other way round you only can use half of the backbone@s bandwith as 
protected tribs bandwidth. A protection concept based on redundant 
ATM links and PNNI on the other hand would give slower recovery times 
(tear down and reestablishment of SVCs by signalling) without the 
need for a priori planning and thus waisting a lot of backbone 
capacity.

Johannes
---------------------------------------------------------------------------
Pan Dacom GmbH
Training/Consulting
Johannes Krohn
Robert-Bosch-Str. 32
D-63303 Dreieich

Tel.: +49-6103/932-137
Fax: +49-6103/932-400
Email: krohn@ffm.pandacom.de