Last update: Fri Jan 5 02:05:58 MST 2018
@Article{Despres:1991:GHS,
author = "R. Despres",
title = "Global High Speed {WAN} Architecture for the 90's",
journal = j-COMP-NET-ISDN,
volume = "23",
number = "1--3",
pages = "125--128",
month = Nov,
year = "1991",
CODEN = "CNISE9",
ISSN = "0169-7552 (print), 1879-2324 (electronic)",
ISSN-L = "0169-7552",
bibdate = "Sat Sep 25 15:30:02 1999",
bibsource = "ftp://ftp.ira.uka.de/pub/bibliography/Distributed/networks.bib;
http://www.math.utah.edu/pub/tex/bib/compnetisdn.bib",
acknowledgement = ack-nhfb,
journal-URL = "http://www.sciencedirect.com/science/journal/01697552",
memo = "The author argues that CONS (Connection-Oriented
Network Services) should be used in future high speed
networks as they provide flow control. The author
states that CLNS (Connection-Less Network Services),
Frame relay (FR) and Asynchronous Transfer Mode (ATM)
do not provide flow control. The author claims that
X.25 works well at 2Mbit/s and for error prone links is
better than CLNS or FR as it has local rather than
end-to-end error control. As IP has no flow control,
retransmissions of discarded packets can lead to
instability Encapsulating IP over X.25 apparently
prevents instability in IP networks at local sites from
spreading out over the WAN. The author claims that CONS
is suitable at speeds greater than 500Mbit/s because
``most of the high speed traffic comes from bulk data
transfers where user data units are very long''. The
author favours the use of 64Kbyte packets on high speed
networks to increase throughput. The author claims that
X.25's second big plus point is that it also offers
packet reassembly. The author claims that X.25 should
be adapted to future high speed networks by moving its
emphasis to multipoint rather than point to point. The
author states that he feels X.25 and ATM should operate
together in a future network to provide CONS
operation",
references = "5",
where = "None",
}