Transcript - NORDUnet

The NORDUnet Network
25 Years of Nordic Research Networking,
Hørsholm, Tuesday the 24th of May 2005
Markus Sadeniemi
[email protected]
NORDUnet 24 May 2005 1
NORDUnet 24 May 2005 2
New world of possibilities: Nordunet
• Nordunet network operations started towards the
end of year 1988
– initial capacity 64 kbps (56 kbps to US)
– before that a 9.6 kbps EARN line
– (Funet-lines inside Finland 64k, 19.2k or 14.4k)
• 64 kbps was enough, the next upgrade to 128 kbps
was made in January 1991
• start of the Internet connection to US was not
without problems
– political problems: is Finland allowed to join?
– security problems: Lawrence Livermore laboratories
were cracked before the network was inaugurated
– crackers were traced to Finland…
NORDUnet 24 May 2005 3
From a loose consortium to a company
• Nordunet Programme was financed by the Nordic Council
of Ministers and hosted by SICS, the Swedish Institute
for Computer Science
• Nordunet network activity started as a loose consortium
• UNI-C kindly offered to host it
– it was viewed as temporary solution, though
• various organizational models were discussed
– a limited company
– foundation
– consortium
• decision: a limited company
• required government decisions
– legal documents ready in July 1991
– Norwegian government approval November 1991, but
– NORDUnet A/S formed on the 14th of December 1993!
NORDUnet 24 May 2005 4
European contacts
• Europe had recognized the need for a R&E backbone
network
• a 64 kbps network was planned using ISO OSI
standards, X.25 et. al.
• EC funded to a large extent both planning and running
the network
– planning took a long time
– call for tender took a long time
– negotiations with the Commission took a long time
• The Nordunet plug and IP was considered bad
manners (or worse), X.25 was the right thing to do
• "EC can support only ISO standards, supporting IP is
legally impossible."
• Things changed in a few years and EC became a
positive element in European networking
NORDUnet 24 May 2005 5
Technology
• Nordunet has always been pragmatic: use what is available
and affordable
• EARN protocols, DECNET, IP and X.25
• LAN technology in long-distance environment
– that's how Nordunet started, now emerging in a new form
in lambda networks
• land based lines, satellite connections, radio links and now
only fibres
• US connection started as a satellite connection
– long latency: 4*30,000 km, a long way to go even for light
– from satellite to transatlantic fibre when upgraded to 34
Mbps
• Frame Relay, ATM fashionable for a while
• IPv6 endorsed by the Commission
NORDUnet 24 May 2005 6
PTTs for better for worse
• National PTTs still had monopoly, when Nordunet
started
• Sometimes a monopoly-PTT was a good partner
– supported research activities
– had good resources
– had spare line capacity
• Sometimes less good
– FI-PTT: "Measure your traffic, so we can volume
charge you."
– FI-PTT: "You may only run ISO traffic on leased
lines."
– France Telecom of European backbone: "You don't
need 34 bps."
NORDUnet 24 May 2005 7
PTT monopolies brake
• After deregulation it takes a few years for
competition to gain momentum
– at first there are no competitors
– in the beginning they are weak
– (exception: national market in Finland)
– Initially competitors' prices only slightly below PTT
monopoly prices (of course)
• Nordunet line SE-DK in 1997
– Danish deregulation in summer 1996
– 34 Mbps line Stockholm- Copenhagen excessively
pricey
– universities of Lund and Copenhagen wanted a
connection, so did Nordunet
– Zone Systems made a n*155 Mbps radio link over
Öresund for us
NORDUnet 24 May 2005 8
Basic question: what for?
• For the users, of course!
• For research and education
• Universities are the core customers
– both for staff and students
– fairly liberal rules for student usage (usually a
university internal decision)
– internet-minded students pouring out of universities
since 1990
• research organisations, lower level schools etc.
served by national network organisations to a varying
degree
• applications (mainly) invented by users, not by
Nordunet
NORDUnet 24 May 2005 9
Working among friends
• The five partners have always had the common goal:
serving our R&E community
• Nordunet has always been pragmatic
– use the technology that serves our community now
– not religious wars on protocols
• During the years the five have sometimes had different
policies
– what technology to use nationally
– how to cooperate with the rest of Europe
• Totally reliable partners
– agreements always respected
– payments always in time
• Fair decisions based on objective needs, not national
interests
NORDUnet 24 May 2005
10
Eastern Europe
• When the eastern countries regained their freedom,
Nordic countries were quick to react
• National networks helped to build connections to the
research communities in
– Estonia, Latvia and Lithuania
– Russia
– Poland
– Ukraine
• Nordunet started projects to help and gave connection on
favourable terms.
• Nordic Council of Ministers was of great help.
• Often the countries' research communities were not well
organized, which sometimes meant long negotiations with
competing parties.
• Help is not really needed any more, we only cooperate for
mutual benefit!
NORDUnet 24 May 2005
11
Cost crisis
• Nordunet costs are essentially paid by the five
national networks
• Some money from the Commission for the European
backbone
• US contribution in the 90s
– Nordunet got a big share of NSF international
budget, small share of NDN-US costs
• Basic problem in the 90s
– traffic volume increased 150% per year
– line costs decreased by 40% per year
– so we get 2.5*0.6 = 1.5
– Nordunet costs increased by about 50% every year
(almost all of Nordunet costs were line costs)
NORDUnet 24 May 2005
12
Nordunet Net Budget
18
16
14
12
10
1 M€
Net Budget M €
50% per year
8
6
4
2
0
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
year
NORDUnet 24 May 2005
13
Nordunet Net Budget
25
20
15
1 M€
Net Budget M €
50% per year
10
5
0
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
year
NORDUnet 24 May 2005
14
Datacom market
• Datacom market was opened for competition in our
countries mainly during 1992-1996
• Nordunet experienced a yearly 40% drop in
transmission prices
• Writing off a transatlantic connection
– formerly in decades
– now in a few years, because of advances in
technology
• IT bubble burst, too much fibre, too much line
capacity
• In 2000-2002 prices dropped radically
• Major providers filed bankruptcy
• Nordunet service in grave danger but survived
NORDUnet 24 May 2005
15
Networks for what?
• Nordunet was a leading network in Europe for many
years (we still are, but others have been catching up)
• But did we know, what we were building networks
for? Maybe not
– we wanted connection to remote central computers –
we got a mail and network news network
– we built a network for mail exchange – we got a
network for file transfer and ftp services
– we built a network for ftp – and got World Wide
Web!
– we built a network for distance education and other
video applications around 1997 – it's done, but the
traffic volumes have always been only a few percent!
– now we build networks for grids and special
applications …
NORDUnet 24 May 2005
16
So what has happened?
• Line capacity from 64 kbps to 10 Gbps
– increase by a factor of over 150,000
– or 17 doublings in 16 years
• Internet was an exotic development project and is
now a part of everyday life (not only in universities)
• It's so natural part of the environment, that we
don't think of it any more
– mail exchange with colleagues around the world
– access to scientific articles
– finding all kinds of information easily
• 20 years ago these could not be done at a click of a
mouse
NORDUnet 24 May 2005
17
Do we need R&E networks?
• Up to (about) 1996 there were no commercial
Internet services worth mentioning
– we had to provide our own services
• since then people (including me) started to say:
"Within 4 years commercial Internet services
become stable, standardized and economical, we
don't need separate R&E networks any more."
• NSF tried that and it did not work
• Practically every country has a national research and
education network
• NRENs are still in the forefront of network
development
• Commercial services are used, but they don't fulfil all
the needs of the R&E community
NORDUnet 24 May 2005
18