introduction to expres - beyond production e-vlbi services t. charles yun program manager expres...
TRANSCRIPT
Introduction to EXPReS- Beyond production e-VLBI services
T. Charles YunProgram Manager EXPReS Project, JIVE
2006 November 21 IST 2006- Helsinki, Finland Slide #: 2
Presentation Overview• Introductions:
• EXPReS• VLBI• Correlation (analysis)
• Some lessons and thoughts
2006 November 21 IST 2006- Helsinki, Finland Slide #: 3
What is EXPReS?• EXPReS = Express Production Real-time e-VLBI Service
The overall objective of EXPReS is to create a production-level, real-time, “electronic” VLBI (e-VLBI) service, in which the radio telescopes are reliably connected to the central supercomputer at JIVE in the Netherlands, via a high-speed optical-fibre communication network...
- or -
Make e-VLBI routine, reliable and realistic for astronomers
Introduction to EXPReS
2006 November 21 IST 2006- Helsinki, Finland Slide #: 4
EXPReS Details• EXPReS is made possible by the European Commission (DG-
INFSO), Sixth Framework Programme, Contract #026642
• Project Details• Three year, started March 2006• International collaboration• Funded at 3.9 million EUR
• Means: high-speed communication networks operating in real-time and connecting some of the largest and most sensitive radio telescopes on the planet
Introduction to EXPReS
2006 November 21 IST 2006- Helsinki, Finland Slide #: 5
Activities in EXPReS• Networking Activities
• NA1: Management of I3• NA2: EVN-NREN Forum • NA3: e-VLBI Science Forum • NA4: e-VLBI Outreach, Dissemination & Communications
• Specific Service Activities• SA1: Production e-VLBI Service • SA2: Network Provision for a Global e-VLBI Array
• Joint Research Activities• JRA1: Future Arrays of Broadband Radio Telescopes on
Internet Computing
Introduction to EXPReS
2006 November 21 IST 2006- Helsinki, Finland Slide #: 6
EXPReS Partners• Joint Institute for VLBI in Europe (coordinator), the
Netherlands• AARNET Pty Ltd., Australia• ASTRON, the Netherlands• Centro Nacional de Informacion Geografica, Spain• Chalmers Tekniska Hoegskola Aktiebolag, Sweden• Commonwealth Scientific and Industrial Research
Organization (CSIRO), Australia• Cornell University, USA• Delivery of Advanced Network Technology to
Europe Ltd. (DANTE), UK• Instituto Nazionale di Astrofisica, Italy• Instytut Chemii Bioorganicznej PAN, Poland• Max Planck Gesellschaft zur Foerderung der
Wissenschaften e.V., Germany• National Research Foundation, South Africa• Shanghai Astronomical Observatory, Chinese
Academy of Sciences, China• SURFNet b.v., The Netherlands• Teknillinen Korkeakoulu, Finland• The University of Manchester, UK• Universidad de Concepcion, Chile• Uniwersytet Mikolaja Kopernika, Poland• Ventspils Augstskola, Latvia
Introduction to EXPReS
2006 November 21 IST 2006- Helsinki, Finland Slide #: 7
Primer- VLBI• A radio telescope looks at an object in the sky and collects data
to create an “image” of the source• Multiple telescopes can view the same object. The distance
between the telescopes is the baseline. The baseline can be compared to building a single telescope with the diameter of this distance (sort of).
• The resolution increases with additional telescopes and longer baselines
• Correlation is the process by which data from multiple telescopes is collected and processed to create a more accurate image. The correlator a super computer (interferometry)
• The sensitivity of the image increases with the data collection rate at the telescope
Introduction to VLBI
2006 November 21 IST 2006- Helsinki, Finland Slide #: 8
Once upon a time...• Telescopes collected data on
tapes… heavy and bulky… postal mail… once all the tapes arrived… tapes were lost/damaged… hard drive arrays slightly improved the situation...
• It was not unusual for the time between experiment to the beginning of correlation to be multiple weeks.
• Today, you can transport the data over the network:e-VLBI - electronic VLBI
Introduction to VLBI
2006 November 21 IST 2006- Helsinki, Finland Slide #: 9
Why transport data over the network?• Using the network to transport
data improves science• Eliminating the need to move
physical objects enables:• Real time analysis• Ability to identify minor
problems in data collection • Hybrid observations• Responsiveness to transient
events• Automated observation
(hands-off observing)• Networked data supports flexible
analysis
Introduction to VLBI
2006 November 21 IST 2006- Helsinki, Finland Slide #: 10
Primer- Correlation (Analysis)• Synthesis imaging simulates a very large telescope by measuring
Fourier components of sky brightness on each baseline pair• EVN MkIV data processor at JIVE
• custom silicon, 1024 chips• Input data is 1 Gb/s max• Around 100 T-operations/sec
• Dedicated, purpose designed/built hardware
Introduction to Correlation
2006 November 21 IST 2006- Helsinki, Finland Slide #: 11
Once upon a time…• Cost to build correlator…
limited flexibility (majority of computation in custom hardware)… preset data input rates… scheduling of scarce resource (correlator)… upgrade cost forces longer life-cycle than desired
Introduction to Correlation
2006 November 21 IST 2006- Helsinki, Finland Slide #: 12
Why “Grid-ify” correlation?• Grid computing offers promising
possibilities:• keep up with input (e.g.,
LOFAR on BlueGene)• Higher precision and new
applications• Better sensitivity,
interference mitigation, spacecraft navigation
• Can CPU cycles be found on the Grid?
• From 16 antenna @ 1Gb/s (eVLBI) To 1000s at 100 Gb/s (SKA)
Introduction to Correlation
2006 November 21 IST 2006- Helsinki, Finland Slide #: 13
Lessons LearnedEach of these bullets is a set of papers, posters and presentations in and of itself…
• Networking is coordination• EXPReS participants on 6 continents
• Connectivity• Networking assumes connectivity, Last mile issues
• Saturating the network is hard• End host hardware• End-to-end Network optimization
• Designing new applications• Custom software- operational vs. proof of concept • Flexible solutions- address current problems, future needs
Reflection.
2006 November 21 IST 2006- Helsinki, Finland Slide #: 14
Looking Forward• Much has been done before
• Importance of standards, open source• Look at other leaders in the field
• Collaboration• Working across disciplines, continents• Partnering to fill gaps (e.g., cpu hardware, analysis
algorithms, visualization, network, storage)• Shared investments
Reflection.
2006 November 21 IST 2006- Helsinki, Finland Slide #: 15
Questions/Answers• Contact information
T. Charles YunProject ManagerEXPReS (JIVE)tcyun \at\ jive dot nl
• Additional Informationhttp://expres-eu.org/ [note: only one “s”]http://www.jive.nl/
• EXPReS is made possible through the support of the European Commission (DG-INFSO), Sixth Framework Programme, Contract #026642
Conclusion
2006 November 21 IST 2006- Helsinki, Finland Slide #: 16