PolyLAB: a new facility for the new way to work.

PolyLAB is a Sun Technology and Research Excellence Centre dedicated to network computing, from collaborative Java-based networks, to wireless and space-based communication systems.

PDG has been given an academic equipment grant by Sun Microsystems  to build the lab, which is thus unique in being dedicated to high-powered communication, instead of high-powered computation.

Present Specifications:


 
Two large servers, Enterprise 450 class.
First 4 x 300 Mhz CPU, 1 Gigabyte RAM, and second 2 x 300 Mhz CPU, 0.5 Gigabyte RAM. Both with 32 Gigabytes of disk space, 40 Gigabyte DLT4000 backup each. Both with 100 Mbps ethernet.

The first E450  (Kasei) has a 155 Mbps Fibre Optic ATM card. Acts as primary machine for delivering content, including full streamed Audio and Video, at production quality.

The second E450 (Nirgal) acts has a host for a range of development environments, and provides support for networking computing, like:
 
 
Four next-generation Javastations
Each one has 64 Megs of RAM, 8 Megs of Flash-RAM, and 100 Mbps ethernet. Runs 100% in Java, and will act as development and test environment.

The result is one of the most powerful facility in the world for investigating the use of computers for scientific and engineering collaboration over the network.

What do we do with it?!

Investigate new ways to communication in all the areas of technical expertise. Explore new communication mediums, and push the edge of technology.

Directions of Networking

We are all getting very used to present Internet hardware. It tends to have lots of `wire' running around (copper or fibre optic) our buildings, and then running off to larger landlines running around the country. We all share the bandwidth of each line, severely limiting network performance.

It is possible to use ATM to control the quality of service (QoS) of a line: can dedicate part of the bandwidth for each user and service. Part of our work is investigating how this may benefit us when we're trying to use the network for intense collaborations.

However, the machines covered by networking goes roughly as the square root of the amount of fibre run (if we wish to maintain bandwidth). Running full-scale ATM around the world is almost impossible!

But fully-interactive communication, using audio, video, and distributed computing, needs high-quality connections!

Does that mean all this is impossible?

No: Move the network layer into orbit.

This is seriously being done. Possible to deliver high-speed networking from orbit, with full 2-way, T-1 and above, connections. Can be done cheaply, and efficiently. Large Comsats's in Geostationary orbit can deliver full ATM to quite small Earth Terminals. There have already been projects by other teams that link Crays together using Craylink, and the ACTS Comsat.

New generation of Low Earth Orbit SatComs:

TeleDesic: A system from Boeing and Microsoft, with 288 SatComs. To be completed in 2002.

Iridium: 66 SatComs, due to be online this year.

These systems can offer very high performance, with very small hardware.

PolyLAB and SatCom

          
The future of the network is almost certainly space-based, but that greatly changes the way in which it works.

The PolyLAB wants to deliver TeleLearning, and TeleMedicine, to everybody in the world. This means working with the networks we can expect.

The people who need the applications are often the ones that cannot get landline connectivity: remote places, ships, to scientists and engineers working in the field and emergency response personel, and even on board the space station.

Problems of SatCom

High latency: high data rates, but long time to get there (especially with GEO).

ATM protocols: Need to fully implement QoS controls.

Uses

People need to be able to collaborate from some pretty remote locations!

The PolyLAB will support the work of the PDG in building fully collaborative networks that can deliver to a globally distributed population, and provide for the needs of scientists and engineers everywhere. The largest machine, Kasei, which has direct backbone connectivity, will be able access the following next-generation networking solutions:

Primary uses will be:

We do this by looking at how to place network computers anywhere, by providing communications infrastructure that allows them to boot and communicate with global networks. We need to support mobile computing, and access to sophisticated services running on supercomputers. Much of this will be done using technology developed by the PolyMath Development Group.  
 


SPB