ÐÓ°ÉÔ­´´

Every day, the space science department of the Rutherford Appleton Laboratories
near Oxford collects massive quantities of raw data from sensing satellites
– enough to fill 2500 floppy discs. The data, in the form of infrared images
of the Earth’s surface, is used to detect changes in sea surface temperatures
for research into climate change. By the time this data has been processed
into a form that makes it useful for scientists, it will have expanded to
fill the equivalent of 6000 discs.

Until now, scientists who wanted to access this data had to either
go to RAL, or wait for magnetic tapes to arrive through the post. Soon,
using the SuperJanet high-speed academic network, they can study the data
in the comfort of their own laboratories, compare it with information from
other satellites on the same screen, and share it with colleagues at other
centres linked to the network.

Chris Mutlow, of the space science department at RAL, says SuperJanet
will change the way he and his colleagues work. ‘We will be able to have
a large archive of processed images which people will be able to browse
through, instead of having to move large quantities of data around on tape,’
he says.

Next week (November 10) is the first anniversary of the launch of SuperJanet
when the Universities Funding Council (now replaced by the Higher Education
Funding Councils of England, Scotland and Wales) gave the go-ahead for the
network to be developed, and awarded an £18 million contract to BT
over four years to build it. A further £2 million was awarded to
the academic community to pay for the equipment they needed to plug into
the network.

SuperJanet is an extension of Janet, the Joint Academic Network, which
has linked 200 higher education institutions since 1983, allowing academics
around the country to communicate by computer – sending electronic mail
messages, publishing papers, and exchanging information with overseas colleagues
through Internet, the world’s largest open access computer network.

What is special about SuperJanet is that it can carry up to 1000 million
bits of information per second – a thousand times more than Janet. With
SuperJanet it would take half a second to send the Complete Works of Shakespeare
from one point on the network to another site 250 kilometres away. The new
network is also well suited to carrying audio and video, as well as text
and graphics. It is actually two networks: a high-speed data network, and
a network using data transmission technology known as asynchronous transfer
mode (ATM), which is better at handling audio and video than conventional
technologies such as packet switching, used by the telephone networks .

SuperJanet’s higher speeds, together with its ability to transmit video
and audio, mean that a whole new set of app-lications, which were not possible
using Janet, can be run over the network. In the year since SuperJanet was
launched, pathologists have exchanged images of tissue samples over the
network, rare documents, such as 13th-century Islamic manuscripts, have
been viewed simultaneously by specialists at different locations, and scientific
papers, containing moving images, have been published and distributed using
SuperJanet.

Next week, a conference will be held in London to demonstrate medical
applications on SuperJanet. These will include using live video links to
teach surgery, and anaesthesia, where a senior anaesthetist can watch and
advise a trainee working in another part of the country.

But the first year has not all been plain sailing. One of the problems
faced by the Joint Network Team, which manages the Janet network and is
responsible for setting up SuperJanet, was deciding which institutes should
be connected, and in what order. The original plan was for a pilot network
of six sites to be set up in March, with six more sites to be added to the
pilot network in September, and at least 29 additional sites to be connected
in November. This has now been revised. Only the 12-site pilot network has
been completed, and according to Bob Cooper, director of networking at the
Joint Network Team, the new target is to have 55 sites linked to SuperJanet
by March.

Not all those sites will have access to ATM technology immediately.
So far, four of the 12 sites are linked to the ATM network, and that will
extend to all 12 sites within the next year. Eventually, the two networks
will be merged, but until then all the other sites will be connected to
the data network only, using fibre-optic lines which can carry data at a
speed of 10 megabits per second. This is five times faster than Janet, which
uses a mixture of twisted pair, coaxial cable and fibre-optic lines leased
from BT, Mercury and cable TV companies. The 12 SuperJanet pilot sites each
have four 34 megabits per second lines, two of which can be linked to the
data network and the other two to the ATM network.

Competition between universities for connection to the pilot network
was fierce. They were asked to submit proposals for applications they would
run under SuperJanet and were selected on the strength of those proposals.
The first six successful applicants were at the universities of Cambridge,
Edinburgh, Manchester, Imperial College, London, RAL and University College
London.

The setting up of this six site pilot network was plagued with a number
of political and technical hitches, such as the problem with a demonstration
application designed to link chemists at Imperial College and at Cambridge.
The link was supposed to enable them to work together by sending 3D molecular
models over the network. But delays in obtaining the necessary routing equipment
meant that Cambridge was not linked to SuperJanet until June, instead of
March, as originally intended. The university managed to set up a hurried
demonstration in time for a visit from the Duke of Edinburgh later that
month, but as soon as he had gone, the equipment was dismantled. And Imperial
College is still waiting for Cambridge to install the right equipment so
the chemists can work together.

To take full advantage of SuperJanet, sites such as Cambridge need to
have high-speed fibre-optic networks connecting machines locally through
local area networks, which conform to a networking standard called Fibre
Distributed Data Interface (FDDI). Special network interface cards need
to be plugged into computers to allow them to connect into the fibre-optic
network. And to make full use of the image, video and audio information
that can be transmitted over Super-Janet, the computers on the researchers’
desks need to run high quality graphics software, and include sound cards.
Many universities, including Cambridge, have not yet got this equipment
because it is expensive and time-consuming to install. Although the Universities
Funding Council provided some money for this kind of equipment, it was not
enough to cover the cost.

A new way to work

Henry Rzepa, a reader in chemistry at Imperial College, is still excited
by the possibilities of the molecular modelling project, and the network
as a whole. ‘We imagine it will be a completely new way of chemists working.
We see it as a replacement for the fax. The fax is error prone and comes
to the recipient dead. You can’t process it. All you can do is look at it
and copy it by hand. With SuperJanet the information is live, in colour
and three-dimensional,’ he says.

Rzepa believes SuperJanet will make a huge difference to groups of scientists
working together. ‘The way science is going is to have pools of people
who can combine their resources. It’s more and more difficult to find the
expertise you need within your own department.’

The broad spectrum of applications outlined in the proposals surprised
the Joint Network Team. ‘When we started a lot of people said that high-performance
networks were for a few select areas, such as supercomputing. That is what
has come out of the US National Research and Education Network. But with
SuperJanet we’ve learnt that such networks are useful to all researchers
and to the population at large. The mass-market appeal of some of the applications
convinced me that we should have a pervasive network. There was a discussion
about whether it should go to a few places or a lot,’ says Cooper.

Although Janet is an academic network, non-academic organisations can
apply to join it, if they are prepared to pay for the service. ICI’s research
laboratories, as well as a number of academic publishers, are linked to
Janet. The same rules will apply to SuperJanet, but as Cooper points out,
outside organisations will not pay to connect to SuperJanet unless it is
widespread enough for them to get some benefit from it. Issues such as how
much will be charged for connection to SuperJanet, and who will be allowed
in, have not yet been resolved. Funding for SuperJanet continues until 1997,
when BT’s contract comes to an end. What happens after 1997 has yet to be
decided.

One organisation which has a keen interest in the potential of SuperJanet
is the NHS. Richard Wootton, of the Royal Postgraduate Medical School at
Hammersmith Hospital, London, which is connected to the pilot network through
a fibre-optic link to Imperial College, is the prime mover behind the medical
applications conference this week. One of his aims in setting it up was
to try and open up a discussion on how SuperJanet could be used in the NHS,
and how it would be funded.

‘The technology has implications for the NHS, but there are a lot of
problems associated with it because SuperJanet is an academic network funded
by the Department of Education,’ says Wootton. ‘If we start running NHS
applications over it I’m not sure how the Department of Health will pay
for it. One of the purposes of the conference is to open up the delicate
discussion of how this would work.’

Another problem would be the reliability of the network. If a senior
consultant is guiding a junior colleague through an operation using a video
link, the users have to be certain that the network would not crash, leaving
the junior doctor and the patient stranded. This level of reliability is
not necessary for researchers exchanging information over the network so
it would be unrealistic to bring the whole network up to the standards needed
by the NHS. Wootton believes both these problems could be resolved by allocating
a permanent private circuit on the network to the NHS, something which the
SuperJanet ATM technology would allow.

Links across the seas

As well as giving organisations such as the NHS access to SuperJanet,
the Joint Network Team needs to find a way of providing overseas links for
the network. If its potential is to be fully exploited, researchers must
be able to collaborate with colleagues overseas as well as in Britain. The
extent to which academics in Britain communicate with overseas colleagues
is shown by the fact that some 25 per cent of the traffic on Janet goes
overseas. At the moment, SuperJanet is linked to Internet’s international
research network, but Internet itself is too slow to take advantage of SuperJanet’s
high speeds, and there are no plans to upgrade all of Internet.

The other option is to set up high-speed links to other high-performance
networks such as the US National Research and Education Network first proposed
in 1991. NREN is a key part of the ‘data superhighways’ policy that President
Bill Clinton wooed the voters with in his election campaign. The network,
which is administered by the US National Science Foundation, builds on US-based
Internet connections and the NSFnet (the National Science Foundation’s own
network). These networks are being improved through the greater use of fibre-optic
connections, and the extension of the network to include schools, libraries
and local government.

Cooper’s team is keen to get a direct link into this growing network.
‘There is a lot of pressure for improved international links,’ says Cooper.
‘That is a challenge because international communications is expensive.’
In particular, a direct link would involve laying specialised transatlantic
connections.

But the success of SuperJanet ultimately depends on widespread access.
As Rzepa points out: ‘If the costs (of connection) are exorbitant, it will
kill the concept stone dead. People may continue to use the phone and fax
instead. You will be able to communicate with people with expertise and
money, but you may want to work with people with expertise and no money.’

Jenny Mill is a freelance writer specialising in information technology.

* * *

Clearing the way for video

Although only a handful of SuperJanet sites have ATM connections so
far, the ATM section of the network is where the future lies because of
its ability to transmit sound and video as well as data. These different
types of information can be combined in many different ways.

ATM was conceived in the early 1980s as a technology for switched public
telephone networks. It is a way of breaking up data into 53 byte cells,
or packets, and sending them over a network by using a series of switches
to direct them from place to place.

With existing packet-switching technology, data is broken up into packets
of different sizes, which means it is more complex to route them around
the network. Software is required to analyse each packet, find out how big
it is, and then decide on the most efficient route. Because all ATM cells
are the same size, it means that they do not need software to route them
around the network. Instead, they can be switched by hardware embedded in
silicon chips, which is much faster because it is a much simpler process
than software routing.

The size of the ATM cells was chosen as a compromise between the needs
of voice traffic, which requires shorter cells to avoid the problem of time
delays that can cause echoes, and data which requires longer ones. The
compromise means that ATM can handle both types of traffic.

With ATM, the data can be sent from a personal computer linked to a
local area network to a high-speed ‘backbone’ network like SuperJanet, and
then to another local area network, using the same switching technology.
To take full advantage of ATM speeds, the local area networks need to use
ATM technology as well, but local area networks using existing technologies,
such as Ethernet or Fibre Distributed Data Interface, can be linked to ATM
networks using special routing equipment. Ethernet and FDDI send all their
messages down a single, shared line so all the messages compete for space
on the line. In contrast, the switching that takes place over ATM networks
means that all connections are effectively ‘private’, just like two people
speaking over a telephone line.

But unlike other switching technologies, ATM uses a concept called ‘virtual
paths’ to allocate space on the network. This concept allows network operators
to allocate fixed chunks of bandwidth, or paths, between nodes on the network.
The bandwidth can be divided up in any way, but once the virtual paths have
been allocated, no other traffic can use them. The paths can then be divide
up further into virtual circuits. And each can have as many virtual paths
as its physical connection to the network allows.

For example, if the total bandwidth is 155 megabits per second, site
A can allocate 15.2 Mbits per second for a video connection to site B, 20.98
Mbits per second for two video links to site C, 15.4 Mbits per second for
a link to the data network, and still have plenty left over. This means
that video, which eats up huge amounts of bandwidth, can be sent without
competing with other traffic on the network and slowing everything down.

But ATM is still in its early days. As the Joint Network Team’s Bob
Cooper says: ‘We’re going to have technical problems with the ATM network
because it is new technology. We will overcome the problems jointly with
µþ°Õ.’