UKLight Application Areas Particle Physics and Radio Astronomy

1
UKLight Application AreasParticle Physics and
Radio Astronomy

• Nicola Pezzi

2
Summary

• CDF
• eVLBI

3
CDF Brief Overview

• Collider Detector at Fermilab
• PP experiment based at FermiLab
• Involves 800 physicists (50 in UK)
• Data stored in the US
• 1000TB per year
• 2000TB stored so far
• 10,000TB predicted for 2008

4
CDF UK Role

• Get local copy datasets (0.5-10TB)
• re-process with better calibration
• Share data/results with other UK physicists and
  CDF physicist worldwide
• UK sites include UCL, Oxford, Liverpool, Glasgow

5
CDF Ultimate goal

• Send data from FNAL to multiple UK sites using
  StarLight/UKLight
• Reprocess data in multiple UK sites and have
  results transparently visible to all CDF users
• Transfer data back to FNAL and update catalogue

6
CDF Issues (1/2)

• Reliable high speed UK-US connection currently
  1GE between FermiLab and UCL using UKLight
• Reliable high speed links inside UK
• Access to large CPU clusters
• Disk storage (10TB)
• Software installation

7
CDF Issues (2/2)

• Standard protocols for secure file transfer to
  satisfy local sites and FNAL security
  restrictions(Kerberos5)
• A transparent file catalogue system to keep track
  of different copies of the same datasets
• Support for new Grid protocols (EU vs. US
  standards, Grid3 vs. LCG) since it affects sw for
  job submission and file catalogue

8
CDF Plans

• Use of UCL PC farm and later on GridPP/Atlas
  Tier-2 centres
• Modification to CDF software to be more Grid
  enabled (e.g. file transfer using GridFTP)
• Develop an interface to LCG to make use of UK CPU
  resources

9
CDF Data rates so far

• Inside Fermilab 64Mbit/s
• Fermilab -gt UCL 5Mbit/s (JANET)
• Fermilab -gt UCL n.a. (UKLight)
• Issues
• Routing to end hosts OK
• Accounts at FermiLab OK
• Routing to Kerberos servers In progress

10
CDF Connection UCL-FNAL
11
Radio Astronomy - overview

• Join Institute for Very Long Baseline
  Interferometry in Europe
• http//www.jive.nl
• Benefits
• Sees through the earths atmosphere
• Sees through interstellar dust
• Higher resolution than other wavebands

12
Radio Astronomy VLBI Pics
13
Radio Astronomy VLBI Pics
14
Radio Astronomy VLBI Pics
15
Radio Astronomy VLBI Pics
16
Radio Astronomy VLBI ops

• Scientists submit proposals to a review
  committee.
• About twenty experiments selected.
• Three sessions per year, two weeks of continuous
  observations
• Up to 20 telescopes per experiment

17
Radio Astronomy VLBI Storage

• Signals from space are sampled and digitised at
  up to 1Gbit/s
• Magnetic tape (up to 1TB each) are shipped to
  data processor for correlation
• Hundreds of tapes per session

18
Radio Astronomy VLBI Problems

• Tape recorders use a 1980s technology and cost
  100.000 each
• Each tape costs 1200 and can be used only 30
  times
• Record/replay heads cost 9000 each and wear out
  in 1 or 2 years
• No further increase in performance feasible.

19
Radio Astronomy VLBI Correlator

• Highly dedicated super-computer
• 16 thousand billion operations per second
• 1024 correlator chips working in parallel
• Capable of processing 16 X 1Gbit/s data streams

20
Radio Astronomy - eVLBI

• Connecting to the Telescopes is not easy.
• They are deliberately built far from cities so
  its expensive
• The principle must be demonstrated first
• eVLBI will connect up to 6 Telescope to JIVE
  in real-time for 2 years as a proof of concept

21
Radio Astronomy eVLBI partners

• DANTE/GÉANT Pan-European Network
• SURFnet Dutch NREN
• GARR Italian NREN
• UKERNA UK NREN
• PSNC Polish NREN
• DFN German NREN
• KTHNOC/NORDUnet Nordic NREN
• Manchester University Network application
  software
• JIVE EVN Correlator
• Westerbork telescope Netherlands
• Onsala Space Observatory Sweden
• MRO Finland
• MPIfR Germany
• Jodrell Bank UK
• TCfA Poland
• CNR IRA Italy

22
Radio Astronomy eVLBI Targets (1/2)

• Evaluate feasibility of eVLBI
• Interfaces network, correlator
• Network installation cost and timescale
• Data rates
• Improved reliability
• Standards and new capabilities
• Network protocols
• Higher data rates
• Parameter tuning for high data rates

23
Radio Astronomy eVLBI Targets (2/2)

• To see significant network usage with multiple
  Gbit streams converging on JIVE from diverse
  European sources
• Must be seen to enable new science and not just
  solve an existing data-transport problem
• Real-time operation is seen as the ultimate aim,
  buffered operation accepted only as a development
  stage.

24
Radio Astronomy eVLBI Today

• 6 telescopes connected to correlator
• Tape recorders being replaced by PCs with
  multiple hard disks
• UDP is used for the PoC
• US partners are developing a version of RTP for
  eVLBI

25
Radio Astronomy eVLBI early success

• On 28th April first real-time elaboration without
  passing via disks
• 2 hours of full real-time computation were
  performed from 6 sites (Jodrell Bank in the UK)
• Data rate utilised was 32Mbit/s (JANET)
• Hoping for 512Mbit/s by the end of year