Cell Relay Archive

[jbasrur@sahana.bellcore.com (33150-basrur)]

> as long as you only think of pure CES for full E1/T1 links there
> certainly is no advantage in using ATM CES over SDH/SONET. The first
> advantage of using ATM takes effect if you start using channelized
> CES for partly used E1/T1 circuits. Then you might need lesss
> bandwith on the STS level compared to using full DS1 mappings in the
> SONET frame. The bandwidth argument gets more important when
> thinking of voice services (compressed, silence removal etc) that
> can be trunked over ATM VCCs using the AAL-2 taking full
> advantage of the VBR character of the compressed  voice data. In this 
> scenario you can support a large number of voice connections over 
> only a fraction of the full STS-3c bandwidth you might have needed 
> using standard DS1 to STS-1 mappings. While taking advantage out of 
> this bandwidth reduction you still can guarantee a certain quality of 
> service for the voice connections in terms of delay (major impact 
> on talker/listener echo!) and degradations due to loss of data in the 
> network. This is even valid for service over ATM in MANs or WANs 
> where is gets really difficult to guarante limited frame delays for 
> IP datagrams when using VoIP.

Isnt compression of voice done in VoIP ?  In this case, the amount of bandwidth used 
will probably be as low as in the case of AAL2 multiplexing.

However, Unless IP gets a large numbers of voice packets IP is going to introduce a lot 
of overhead...right ? 
Is there a standard as to how voice circuits are put onto IP.?

Okay, if IP is not going to be able to support QoS, what about RSVP ?
Arent people considering using IPv6 with RSVP instead of IP/ATM/SONET.?

Jyotin
> 
>     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
>