Lighting the Blue Touchpaper for UK e-Science - Closing Conference of ESLEA Project

26 March, 07 12:00 PM - 28 March, 07 01:30 PM

The George Hotel, 19-21 George Street, Edinburgh, EH2 2PB

Organiser(s): Colin Greenwood, (National e-Science Centre), Peter Clarke (National e-Science Centre) and Matthew Strong (University of Manchester)

Any slides or other material generated as a result of this event can be found at: www.nesc.ac.uk/action/esi/contribution.cfm?Title=748

Keynote Speakers

Title: tbc

Author: Eric Boyd

Affiliation: Internet2

Format: Keynote Presentation

tbc.

Title: tbc

Author: Edward Seidel

Affiliation: Center for Computation & Technology, Louisiana State University

Format: Keynote Presentation

Professor Edward Seidel

Edward Seidel is the director of the Center for Computation & Technology at Louisiana State University and the Floating Point Systems Professor in LSU’s Departments of Physics and Astronomy, and Computer Science. Seidel is well known for his work on numerical relativity and black holes, as well as in high–performance and grid computing. He earned his Ph.D. from Yale University in relativistic astrophysics. He headed the numerical relativity group as a professor at the Max-Planck-Institut fuer Gravitationsphysik (Albert-Einstein-Institute) in Germany from 1996 - 2003, where he maintains an affiliation. He was previously a senior research scientist at the National Center for Supercomputing Applications and associate professor in the Physics Department at the University of Illinois.

Seidel is a recipient of the 1998 Heinz-Billing-Preis of the Max-Planck-Gesellschaft; a recipient of the 2001 Gordon Bell Prize; and winner of various HPC Challenge awards at SC98, SC01 and SC02. In 2004, Seidel was named the Rising Star of the Year at the Governor’s Technology Awards in Louisiana and one of HPCwire’s Top People and Organizations to Watch. In 2006 he was chosen as the recipient of the Sidney Fernbach Memorial Award. Seidel has been the PI or CoPI on large grants in Physics and Computational Science from NSF, DOE, NASA, the German DFN-Verein, and the European Commission, where he led the EU Astrophysics Network and was a leader in the GridLab project. He is the co-chair (emeritus) of the Applications Research Group, Global Grid Forum and chief scientist for the Louisiana Optical Network Initiative (LONI). He is the author or co-author of more than 150 publications and serves on numerous national and international committees and advisory boards.

Title: Recent developments in Lambda networking

Author: Cees de Laat

Affiliation: University of Amsterdam

Format: Keynote Presentation

This talk will focus on recent developments in Lambda networking. The recently started projects Phosphorus and StarPlane will be presented. It will delve into some studies with respect to cross domain authorization and topology information distribution to facilitate Lambda and LightPath setup. The token based authorization and Network Description Language demonstrators shown at SC06 will be explained.

Dr. Ir. Cees de Laat

Cees de Laat is associate professor in the Informatics Institute at the University of Amsterdam. Current research in his group includes optical/switched networking to optimize Internet transport of massive amounts of data, distributed cross organization Authorization architectures, grid Workflow systems and systems security. With SURFnet he implements projects in the GigaPort Research on Networks. He collaborates in the NSF- OptIPuter project. He serves as Grid Forum Steering Group (GFSG) Infrastructure Area Director and IETF Liaison. He is co-founder and organizer of several of the past meetings of the Global Lambda Integrated Facility (GLIF). < http://www.science.uva.nl/~delaat>

Session 1: Protocols

Title: TCPdelay

Author: Richard Hughes-Jones and Stephen Kershaw

Affiliation: The University of Manchester

Format: Presentation

Transmission Control Protocol (TCP) is a reliable transport protocol which guarantees that data sent will be perfectly replicated at the receiver. In order to deliver on this guarantee, TCP transmits data according to well defined rules which create reliable transfers and attempt to ensure that our traffic can fairly co-exist with other data flows. This can result in large transit delay and low throughput.

We investigate the use of TCP for the transfer of real-time constant bit-rate data and report on the effects of the protocol on the flow of data. We discuss the requirements for TCP to be successfully applied in this situation and the implications for applications such as eVLBI, where we are more concerned with timely arrival of data than guaranteed delivery.

Title: Implementing DCCP: differences from TCP and UDP.

Author: Andrea Bittau and Mark Handley

Affiliation: University College London

Format: Presentation

In this talk we will describe our experiences in implementing a new protocol, DCCP, in the Linux kernel. Being the first implementation in a main-stream operating system, we are the first ones to explore the implications and the unexpected issues that could arise from developing this protocol. We will focus on how the DCCP implementation differs from that of TCP and the performance issues that we have encountered.

We found that DCCP's code complexity approaches that of TCP. We also noted that the two protocols, albeit being very different, may indeed share code regarding congestion control algorithms. The current performance bottleneck is checksum calculation. DCCP allows the calculation of partial checksums and this may become a big win against TCP when sending large packets at high rates.

Title: DCCP

Author: Richard Hughes-Jones and Stephen Kershaw

Affiliation: The University of Manchester

Format: Presentation

Datagram Congestion Control Protocol (DCCP) is a new transport protocol whose development and implementation in Linux is being aided by the work of Mark Handley and Andrea Bittau in the ESLEA project. The protocol is attractive to many applications where data is transferred with tight constraints on the timing of data delivery, such as internet telephony and e-Science applications such as eVLBI.

We investigate the current implementation of DCCP in Linux and report results of performance tests, with discussion of what DCCP can currently do for us and a suggested approach for DCCP options specific for VLBI.

Title: TCP performance

Author: Richard Hughes-Jones and Stephen Kershaw

Affiliation: The University of Manchester

Format: Presentation

Transmission Control Protocol (TCP) has been used in the same basic form for over 25 years. The reliability of data transfer has always been a feature and the implementation of congestion avoidance has long been credited with keeping the Internet operational despite huge increases in traffic, in addition to allowing many streams of data to co-exist in a relatively fair manner. The performance of TCP is governed largely by these features. However, standard TCP is not very suitable for high-bandwidth, long distance networks, and recent developments seek to address this deficiency.

We investigate the alternative TCP stacks available in modern Linux kernels and compare their performance to the standard Reno TCP. We comment on the features of these modern variants and their behaviour when there is congestion, cross-traffic and reverse flows.

Session 2: Protocols and Performance Testing

Title: Utilising UDT to push the Bandwidth envelope

Author: Brian Davies* and Barnaby Garrett#

Affiliation: * Lancaster University; # The University of Edinburgh

Format: Presentation

eScience applications, in particular High Energy Physics, often involve large amounts of data and/or computing and often require secure resource sharing across organizational boundaries, and are thus not easily handled by today's networking infrastructures. By utilising the switched lightpath connections provided by the UKLight network it has been possible to research alternative methods of data transport. While the HEP projects make use of a number of middleware solutions for data storage and transport they all rely on GridFTP for WAN transport. The GridFTP protocol runs over TCP as the layer 3 protocol by default, however with the latest released of the Globus toolkit it is possible to utilise alternate protocols at the layer 3 level. One of the alternatives is a reliable version of UDP called UDT. This report presents the results of the tests measuring the performance of single-threaded file transfers using GridFTP running over both TCP and the UDT protocol.

Title: Trans-Atlantic UDP and TCP Network Tests

Author: Matt Strong, Anthony Rushton, Richard Hughes-Jones, Ralph Spencer, Simon Casey, Paul Burgess

Affiliation: The University of Manchester

Format: Presentation

We report on UDP and TCP network tests performed between Manchester, UK, and Chicago, USA, utilising the UKLight dedicated lightpath between these two endpoints, obtained as part of the ESLEA project. We show near line rate instantaneous UDP throughput over this lightpath, and iPerf TCP bandwidths of in excess of 900 Mbit/s. We also report on network tests between Jodrell Bank Observatory, UK, and Haystack Observatory, USA, utilising the Bossnet network infrastructure and the UKLight dedicated lightpath. Again, we show near line-rate instantaneous UDP throughput and iPerf bandwidths of in excess of 900 Mbit/s.

Title: Performance testing of SRM and FTS between Lightpath Connected Storage Elements

Author: Brain Davies and Roger Jones

Affiliation: Lancaster University

Format: Presentation

We describe the configuration, testing and optimisation of file transfers between LCG middleware and SRM storage systems between two LCG sites using a UKLight connection. We will also discuss recommendations for continued work.

Title: 10 Gigabit Ethernet

Author: Richard Hughes-Jones and Stephen Kershaw

Affiliation: The University of Manchester

Format: Presentation

Network technology is always moving forward and with the recent availabilty of 10 Gigabit Ethernet (10GE) hardware we have a standard technology that can compete in terms of speed with core or backbone Internet connections. This technology can help deliver high-speed data to the end-user but will systems that are currently used with Gigabit Ethernet deliver with 10GE?

We investigate the performance of 10 Gigabit Ethernet network interface cards in modern server quality PC systems. We report on the achievable throughput and comment on the performance of transport protocols at this higher speed.

Session 3: RealityGrid

Title: Application Based Network Performance Testing

Author: Stephen Pickles and Robin Pinning

Affiliation: The University of Manchester

Format: Presentation

The large scale modelling of many physical phenomena increasingly requires the model to be of a size that is too large for one HPC resource. MPICH-G2 is a grid-enabled MPI implementation that allows the coupling of multiple machines for the running of a single MPI-based application.

This work aims to show how the use of light-switched optical networks, such as UKLight, affect codes of this class by running a series of benchmarks using the Intel MPI benchmarking suite.

Title: Large-scale lattice-Boltzmann simulations over lambda networks

Author: R. S. Saksena*, P. V. Coveney*, R. L. Pinning**, S. P. Booth#

Affiliation: * University College London, ** University of Manchester, # The University of Edinburgh

Format: Presentation

Amphiphilic molecules are of immense industrial importance, mainly due to their tendency to align at interfaces in a solution of immiscible species, e.g., oil and water, thereby reducing surface tension.

Depending on the concentration of amphiphiles in the solution, they may assemble into a variety of morphologies, such as lamellae, micelles, sponge and cubic bicontinuous structures exhibiting non-trivial rheological properties. The main objective of this work is to study the rheological properties of sheared gyroidal systems containing defects of very large systems (of up to 1024^3 lattice sites) using the lattice-Boltzmann method. Memory requirements for the simulation of such large lattices exceeds that available to us on most supercomputers and so we use MPIg/MPICH-G2 to perform geographically distributed domain decomposition simulations across HPCx in the UK and TeraGrid in the US. Use of MPIg and MPICH-G2 requires the port-forwarder to work with them on HPCx. Data from the simulations is streamed to the SGI Prism at UCL( London) for rendering and visualisation.

Title: Use of UKLight as a Fast Network for Data Transport from Grid Infrastructures

Author: Mary-Ann Thyveetil, James L. Suter and Peter, V. Coveney

Affiliation: University College London

Format: Presentation

Large-scale molecular dynamics simulations run on high-end supercomputing facilities can generate huge quantities of data. We simulate mineral systems up to 10 million atoms in size in order to extract materials properties otherwise difficult to obtain through existing experimental techniques. Simulating clay systems this large can generate huge files up to 50GB in size.

These simulations were carried out on remote sites across the US TeraGrid and Europe's DEISA grid. Using the dedicated network, UKLight, from these high-end supercomputing resources has significantly reduced the time taken to transport large quantities of data generated from our simulations. UKLight provides excellent quality of service, with reduced packet loss and latency. This enhanced data transfer method paves the way for faster communication between two coupled applications, such as in the case of real-time visualisation or computational steering.

Title: Use Using Lambda Networks to Enhance Performance of Interactive Large Simulations

Author: M Harvey *, S Jha ~, M-A Thyveetil # and P V Coveney #

Affiliation: * Imperial College, London; ~ Louisiana State University; # University College London

Format: Presentation

Session 4: Resource Scheduling and Hosting

Title: The ESLEA Circuit Reservation Software

Author: Clive Davenhall, Peter Clarke and Lihao Liang

Affiliation: National e-Science Centre

Format: Presentation

As part of the ESLEA Project we have developed the CRS (Circuit Reservation Software) which allows dedicated network connections to be reserved or ‘booked’ in advance. The CRS has been deployed to control some circuits running over UKLight which are available within ESLEA. Configuration of the circuits is achieved by having them pass through Cisco 7609 switch/routers placed around the UKLight access points. These switch/routers belong to ESLEA and hence can be configured by the CRS.

The CRS is based on the BAR (Bandwidth Allocation and Reservation) software developed for the EGEE Project. The BAR software supported advanced reservations over a conventional IP network. The BAR architecture has separate components for controlling separate, and administratively distinct, sections of the network. Thus, this architecture is inherently flexible and extensible. It is potentially suitable as a basis for a production service for reserving dedicated circuits. The talk will describe the CRS, discuss the problems encountered in implementing it and consider how it could be enhanced.

Title: Co-allocation of Compute and Network resources using HARC

Author: Jon MacLaren

Affiliation: The University of Manchester

Format: Presentation

HARC - the Highly-Available Resource Co-allocator - is a system for co-allocating resources. HARC can handle multiple types of resource, and has been used to reserve time on supercomputers together with dedicated lightpaths across two continents, as shown in demonstrations at GLIF and SC'06. HARC makes these multiple allocations in a single atomic step; if any resource is not available as required, then nothing is reserved. To achieve this "all or nothing" behaviour, HARC treats the allocation process as a Transaction, and uses a phased-commit protocol. HARC uses the Paxos Commit protocol to ensure that there is no single point of failure in the system, which, if correctly deployed, can have a very long Mean Time To Failure (years).

Here we give an overview of HARC, and explain how the current HARC Network Resource Manager (NRM) works, and is able to set-up and tear- down dedicated lightpaths.

We will also touch upon plans to deploy HARC in the UK, starting with a planned deployment on NW-Grid, and show how a new HARC Resource Manager could be constructed to interface to the ESLEA Circuit Reservation Software.

Title: The Application Hosting Environment: Lightweight Middleware for Grid Based Computational Science

Author: Stefan Zasada

Affiliation: University College London

Format: Presentation

The Application Hosting Environment (AHE) is designed to provide the scientist with a simple, lightweight mechanism for launching and monitoring scientific applications, such as NAMD, LB3D, LAMMPS, DL_POLY, running on remote grid resources. It provides a consistent, centralised way to control application instances running on distributed grid resources, including those provided by the UK NGS and US TeraGrid. The AHE is developed in Perl, and models application instances as stateful WSRF services using the Perl WSRF::Lite toolkit. The AHE is built on top of the GridSAM job submission web service, meaning that it currently provides an interface to grid resources running the Globus Toolkit 2, Unicore, Sun Grid Engine and Condor, and as such provides a uniform interface to resources running these middlewares.

Title: Grid-empowered Optical Burst Switched Network: Architecture, Protocols and Testbed

Author: Georgios Zervas, Reza Nejabati, Dimitra Simeonidou

Affiliation: The University of Essex

Format: Presentation

This paper presents a novel architecture for Grid-enabled optical burst switched networks. We suggest an overall Grid network architecture and we propose a Grid Optical User Network Interface (G.OUNI) able to bridge Grid and OBS network environment. G.OUNI functionalities such as resource availability, discovery, and reservation required to integrate local Grid and OBS domains are presented. Furthermore, extended OBS protocols to support aforementioned services are being described and implemented on an application-aware asynchronous optical burst switched network demonstrator. Fully functional edge and core OBS routers as well as their algorithms and functionalities are being presented.

Session 5: HEP

Title: Configuration of an Endsite to enable Lightpath Capabilities

Author: Brain Davies and Roger Jones

Affiliation: Lancaster University

Format: Presentation

The utilisation and optimisation of lightpath connections with production systems in parallel with production network services can be beneficial to productivity, yet can lead to necessary changes to network configurations. We describe here the benefits, problems and implemented solutions found to enable the Lancaster LCG Tier2 site to operate Tier2 services within a multi-domain network system.

Title: Building a distributed software environment for CDF within the ESLEA framework

Author: V. Bartsch, N. Pezzi and M. Lancaster

Affiliation: University College London

Format: Presentation

The aim of the ESLEA project for CDF is to make computing resources available for users of the CDF collaboration within the UK. The effort has been embedded in the collaboration wide activities to build a distributed computing system called CAF. Software tools common to the collaboration are being used.

For Europe, it turned out that it would be best to engage the CDF computing Grid into the infrastructure of the LHC computing Grid LCG because it is the prevalent Grid computing system within Europe and all computing sites which are considered to run CDF jobs also run an LCG site.

Currently, the UCL HEP site, with a total of 20 CPUs, has been added to the CDF Grid system with two other sites, the UCL CCC site and a site at Liverpool to be added in the near future. A major drawback up to now has been that the storage of data is not yet possible within our framework.

Title: An Extended Storage System across MANs

Author: Brain Davies and Roger Jones

Affiliation: Lancaster University

Format: Presentation

Within the LCG Grid environment it may be beneficial to have singular control management and access to storage devices across multiple sites and across multiple Metropolitan area networks. We describe he the benefits and plans for such a system, progress to date for implementation of an extended dCache and the problems encountered so far.

Title: IS Security in a World of Lightpaths

Author: Robin Tasker

Affiliation: CCLRC, Daresbury Laboratory

Format: Presentation

IS Security is a cornerstone for the delivery of consistent and reliable services in every aspect of the business of an organisation. The traditional IP network service provided to Institutes is carefully managed and controlled to limit illegal and/or antisocial use to protect the business processes of that Institute. SuperJANET5 has the capability for additional bandwidth circuits - lightpaths - to be provided between specific endpoints across the network to meet specific need. Because these are end-to-end circuits they reach right into the heart of an organisation, typically providing a high bandwidth interconnection, and often at rates that are difficult to police. This talk explores this problem space and provides a strategy to minimise any associated risk through the development of an appropriate Security Policy that can sit alongside an Institute's overall approach in this area.

Session 6: VLBI

Title: The Contribution of ESLEA to the Development of e-VLBI

Author: Ralph Spencer*, Richard Hughes-Jones*, Paul Burgess*, Matt Strong*, Simon Casey*, Colin Greenwood#

Affiliation: * The University of Manchester; # National e-Science Centre, Edinburgh

Format: Presentation

e-VLBI - the use of the Internet in real time VLBI high resolution observations in radio astronomy - has become a routinely available technique in this last year. ESLEA has contributed enormously to its development, by improving our understanding of data transmission networks, the limitations of transport protocols and end hosts, and communication of this knowledge to radio astronomers in Europe. A series of tests, organised by ESLEA and by JIVE in the Netherlands has gradually led to open call science runs, now considered as a regular part of European VLBI Operations. A major upgrade project - EXPReS - is now underway to equip more European observatories with e-VLBI capability. This paper outlines the work done in ESLEA on e-VLBI and illustrates its success by showing recently obtained astronomical results.

Title: Investigating the e-VLBI Mark 5 end systems in order to optimise data transfer rates as part of the ESLEA Project

Author: Matt Strong, Ralph Spencer, Richard Hughes-Jones, Simon Casey, Paul Burgess, Stephen Kershaw

Affiliation: The University of Manchester

Format: Presentation

We report on the development of high bandwidth data transfers for e-VLBI at Jodrell Bank Observatory as part of the ESLEA project. ESLEA is a UK project to exploit the use of switched-lightpath optical networks for various applications, including e-VLBI, HEP, High Performance Computing and e-Health. We show how the CPU power of the Jodrell Bank Mark 5 end systems was limiting the data transfer rate to below 512 Mbit/s. Both of the Jodrell Bank Mark 5 end systems have now been upgraded and can now transfer e-VLBI data to JIVE at the required data rate of 512 Mbit/s.

Title: Investigating the effects of missing data upon VLBI correlation using the VLBI_UDP application

Author: Simon Casey*, Ralph Spencer*, Richard Hughes-Jones*, Matt Strong*, Paul Burgess*, Arpad Szomoru~, Colin Greenwood#

Affiliation: * The University of Manchester, ~ The Joint Institute for VLBI in Europe, NL (JIVE), # National e-Science Centre, Edinburgh

Format: Presentation

VLBI_UDP is a data streaming application, specifically orientated at streaming VLBI data. Since the UDP protocol does not guarantee delivery of data, it has been necessary to conduct investigations into how the VLBI correlator deals with portions of missing or corrupt data. The results obtained from these investigations can then be programmed back in to VLBI_UDP to increase reliability of data transmissions. This paper details modifications made to VLBI_UDP to enable dropping of packets within a controlled environment and results from playing the corrupted data in to the correlator. Also presented are examples of streaming data from multiple sites, simulating the network and end-host loading of an e-VLBI experiment.

Title: e-VLBI developments at JIVE

Author: Arpad Szomoru

Affiliation: Joint Institute for VLBI in Europe

Format: Presentation

In this talk I will give an overview of the e-VLBI effort at JIVE over the past few years, with an emphasis on the EXPReS project. I will present the current status of the e-EVN and the developments currently underway.

Session 7: Arts & Humanities

Title: Music and Audio – Oh how they can stress your network

Author: Dr R P Fletcher

Affiliation: The University of York

Format: Presentation

Technology report, “Networking Audio and Music Using Internet2 and Next-Generation Internet Capabilities” (TC-NAS 98/1) by the Audio Engineering Society (AES) made a number of interesting statements (in 1998):

The current Internet is inadequate for transmitting music and professional audio.
Performance and collaboration across a distance stress beyond acceptable bounds the quality of service.
Audio and music provide test cases in which the bounds of a network are quickly reached and through which the defects in a network a readily perceived

Given these key points made nearly 10 years ago, where are we now? Have we started to solve any of the problems from the musicians’ point of view? What does a musician do to cause a network so many problems? A trained musician's ears are extremely sensitive to very subtle shifts in temporal materials and localisation information. A shift of a few milliseconds can cause difficulties.

The sample rate and number of bits needed to represent music in the digital domain has risen over the past few years. 15-20 years ago audio on the net was transmitted using voice compression technologies and around 8k sample rates in mono! This provided truly awful fidelity. In 1992 the mp3 (MPEG 1 Audio Layer 3) format became available and we saw the start of the explosion of the lossy compression technologies and audio streaming technologies. As of February 12th 2007 we now have a MPEG Surround Sound standard.

It might not be unreasonable to require transmission of an uncompressed audio stream running at 192Khz across 16 channels using IEEE 64-bit floating point precision (196,608,000 bits/second). If video is required (and HD is mandatory these days), then additional bandwidth is needed (eg highly compressed at 19.2 Mbps to 1.5 Gbps for raw HDTV)

Collaboration introduces the notion that it is in fact the round trip time which has to be taken into account. What technologies are required to allow remote collaborative composition and performance? High performance computing is also required by musicians for audio processing and production.

Once high performance audio and video can be effectively controlled and transmitted across the networks then a number of avenues open up, e.g broadening musical participation, master classes, remote studio work, collaborative composition/performances, multi-site performances, more timely post processing of film production and so on.

Title: Video-Conferencing from the City Halls, Glasgow

Author: Gill Davies

Affiliation: BBC

Format: Presentation

The City Halls is Glasgow's centre for music performance and education, and provides high quality, diverse and innovative music learning and participation opportunities for people of all ages and abilities.

As part of their learning programme, The BBC Scottish Symphony Orchestra have been conducting a pilot of video-conferenced sessions to schools in remote areas of Scotland including Castlebay School on Barra, Western Isles; Bowmore Primary and Islay High School on Islay, Argyll and Bute; Papdale Primary and Kirkwall Grammar in Kirkwall, Orkney and St Teresa's Primary Dumfries, Dumfries and Galloway.

These schools were chosen as teachers and pupils are experienced users of video-conferencing technology. Their remoteness means video-conferencing is a valued tool in teaching and teacher training.

As the BBC SSO is a touring orchestra, we aim to expand our use of video-conferencing to develop and sustain relationships with schools and communities abroad.

The National Youth Choir of Scotland has used the video-conferencing facilities at the City Halls to audition potential members in remote areas of Scotland.

The City Halls was one of nine centres which took part in a video-conference session with composer John Adams organised by the London Symphony Orchestra. Students from Glasgow University and secondary schools in Glasgow were present in the City Halls and joined students from the four countries in the UK and Eastman School of music in New York.

In my presentation, I aim to talk about how the BBC and associate partners can develop video-conferencing and fully utilise the opportunities afforded by educational networks.

Title: Who "owns" the network: a case study of new media artists' use of high-bandwidth networks

Author: Frédérik Lesage

Affiliation: London School of Economics and Political Sciences

Format: Presentation

Near the end of his classic work on the social construction of art, Howard S. Becker details a fascinating comparison between the development of the stereoscope and the photograph as two somewhat promising artistic applications in the late 19th and early 20th century. In his account, he describes how the latter went on to be accepted by most of society as a tool for artistic creation while the former did not achieve the same level of success. Although Becker is unable to give the reader a definitive answer as to why one succeeded while the other did not, it is fascinating to extend his reflections to new media technologies and to ask how some of these may one day contribute to the work of a contemporary Stieglitz. One approach could involve documenting how these digital technologies are currently employed by artists to gain a wider understanding of the dialogical relationships between artists, the tools they use, and the wider art world that surrounds them.

With this is mind, a case study has been put together to observe how artists apply the network metaphor to information and communication technologies (ICTs), specifically high-bandwidth academic networks, in order to coordinate the production of new media artworks.

Although it is certainly impossible to predict whether contemporary artists' choices of technologies will one day lead to as successful an art world as photography, it is my hope that this research will provide three main points of potential interest to those attending the conference:

How to develop ethnographic tools for the study of organisational networks who use ICTs, particularly those working in the cultural or creative industries.
How a theoretical framework based on the mediation of conventions might lead to a better understanding of the diffusion of creative practices and discourses relating to new media.
An introduction to the activities of the MARCEL Network and a wider understanding of artists' challenges and interests when working with ICTs.

Title: Always the Bridesmaid and never the Bride! Arts, Archaeology and the E-Science Agenda

Author: Vincent Gaffney* and Rob Fletcher#

Affiliation: * The University of Birmingham, # The University of York

Format: Presentation

It appears to be true that the emergence of e-science agendas has not had the impact on Arts and Humanities that it has had on many other disciplines. This is certainly not the consequence of a lack of data for analysis within the Arts. Many Arts subjects have inherited vast amounts of historic data. Others, including archaeology, sit at the edge of the natural sciences and are prodigious generators of data in their own right, or greedy consumers of digital data generated by other disciplines. Neither should the current position be taken as indicative that humanistic enquiry is not amenable to e-research. Whilst it is true that the subjects of Arts enquiry are frequently not mechanistic, in the manner that assists research in many applied sciences, it would be a mistake to assert that Arts research is so chaotic or anarchic that it is impossible to demonstrate the value of e-science applications to our disciplines. This paper will affirm the claim of Arts as data rich environments that can benefit from an e-agenda that goes beyond “resource-discovery” and stake a claim for a greater investment in e-science. In particular, it will consider the requirement of the Arts, and Archaeology, for increasingly sophisticated analysis and representation, ponder upon the current subordinate position of these disciplines and question whether we can actually do any better with the opportunites that exist at the moment.

Session 8: Closing Session

Title: Exploitation of Switched Lightpaths for e-Health: Constraints and Challenges

Author: Lee Momtahan and Andrew Simpson

Affiliation: Oxford University Computing Laboratory

Format: Presentation

We give an overview of our experiences trialling the use of switched lightpaths for e-Health as part of the wider ESLEA project. We identify use cases from a number of projects with the potential to leverage switched lightpaths in the e-Health domain. Consideration of these use cases highlights a wide range of social, legal, ethical and technical constraints which must first be overcome. We document these constraints and detail the challenges we have faced in implementing some of these use cases for a particular e-Health project: the Integrative Biology project.

Title: Evolution Strategies for Next Generation Networks

Author: Dr Craig Michie, Prof Ivan Andonovic, Dr D Harle

Affiliation: University of Strathclyde

The value chain within the communication sector ranges through content/application generation through delivery to consumption. Content is no longer considered in isolation from service provision, service provision in isolation from delivery channels, or delivery channels in isolation from enabling and managing technologies. By implication, technological developments within any one area of the telecommunications chain may have a significant on other components downstream. In recognition of this the Institute of Communications and Signal Processing within Strathclyde has grown a portfolio of activity addressing the needs of a broad range of the communications network including the access network (fixed, mobile and wireless), optical networking; control and management; wireless/mobile heterogeneous networks and wireless sensor networks. This presentation presents an overview of activities within Strathclyde from a network perspective aimed at supporting some of the applications which are relevant to the wider community.

Title: Exludus RepliCator, Virtual Machines and UKLight

Author: Aaron Turner

Affiliation: The University of York

Format: Presentation

A common problem in grid computing is ensuring that a job can run on a wide variety of resources so that scheduling is relative simple and the widest range of resources are available for use. A solution to this is to use virtualisation, but this can involve the transport of large virtual machines and provisioning on a remote resource.

It is proposed to run experiments using RepliCator from eXludus, over both YHMAN and UK Light between Leeds and York to test the viability of the use of virtual machines for grid computing.

RepliCator from eXludus is a data provisioning tool, and the White Rose Grid at York is the European partner for this software. The software provides transparent data provisioning for large data objects by modifying transmission and providing a caching facility.

UK Light is of interest due to the ability to reserve specified levels of bandwidth that might be needed to support any fully operational grid based on the use of virtual machines.

Title: Managed Bandwidth Services in SuperJANET5

Author: David Salmon

Affiliation: UKERNA

The UKLight infrastructure is being migrated to capacity provided within SuperJANET5, and the SuperJANET5 transmission infrastructure itself can also provide high-capacity circuits to projects with a persistent requirements. The presentation will provide an overview of these services and how they are implemented and deployed within JANET, including policies governing use and costs. This should be useful for any of the research groups who have been working within the ESLEA project and who may be considering continuing their work in some form after ESLEA is completed.

Posters:

Title: CARMEN and UKLight

Author: Aaron Turner

Affiliation: The University of York

Format: Poster

Code Analysis, Repository and Modelling for e-Neuroscience (CARMEN) is a £4.5 million project over 5 years led by Colin Ingram at Newcastle.

As part of this project two data and processing centres are being set up, one at Newcastle, one at York. This will comprise large data storage (10TB per site initially, growing to over 50TB in total), shared and replicated via the Storage Request Broker (SRB) and processing clusters to perform data search and analysis, for example spike train analysis from information from sensors monitoring brain activity.

It is proposed to examine the feasibility of utilising a UK Light link between the sites to provide an assured level of data transfer capability. This will then allow a number of novel approaches to be deployed, including: parallel analysis over large datasets; real time analysis of neuroinformatics data streamed from experiments on valuable but short-lifetime tissue; dynamic data integration; high speed transfer and replication of very large data sets for performance and reliability.

Title: Monitoring the UKLight network

Author: Barnaby Garrett

Affiliation: The University of Edinburgh

Format: Poster

The eScience applications taking part in the ESLEA project aim to utilise the switched light paths offered by UKLight to deliver massive datasets and prove bandwidth intensive data streaming technologies. For debugging, analysis and reporting, the applications need to be able to monitor their bandwidth utilisation. This paper looks at the methods that could be used and the implementation of those technologies within the UKLight infrastructure.

PLUS posters from ESLEA and UKERNA tbc

http://www.nesc.ac.uk

The National e-Science Centre