1 / 18

The ITER Data System Challenges

The ITER Data System Challenges. presented by Jo Lister, CRPP-EPFL, Switzerland with Izuru Yonekawa , JAEA-Naka, Japan John How , CEA-Cadarache, France / ITER-IT So Maruyama , ITER-IT, Germany. ITER update Where are ITER’s main data challenges ?

hope
Download Presentation

The ITER Data System Challenges

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. The ITER Data System Challenges presented by Jo Lister, CRPP-EPFL, Switzerland with Izuru Yonekawa, JAEA-Naka, Japan John How, CEA-Cadarache, France / ITER-IT So Maruyama, ITER-IT, Germany • ITER update • Where are ITER’s main data challenges ? • Lead up to today’s round-table discussion

  2. Where will ITER be sited ? • After long negotiations, a European site was agreed at the end of June 2005 • Cadarache (CEA), Bouches du Rhône, France • Commissioning in or after 2016! • Now we have structural discussions • Director general • Staffing structure • Financing • etc etc • ITER parties are still EU+CH, JA, China, Korea, Russian Federation, USA • Long lead item procurement starts as soon as possible • At least 7 year construction • At least 1 year integrated commissioning Jo Lister, CRPP-EPFL, ITER Data Challenges, ICALEPCS 2005, Geneva, October 2005

  3. Reminder of what ITER looks like - physically 28m Fusion Power = 500MW Fusion Power/Auxiliary Heating Power = Q>10 Neutron wall loading = 0.57 MW/m2 Plasma major radius = 6.2 m Plasma minor radius = 2.0 m Plasma Current = 15 Megamp Toroidal Field = 5.3T Plasma Volume = 837 m3 Heating power = 73 MW Pulse lengths = 300 - 5000 sec PF supra coils = 925+6*130-390 tons TF supra coils = 18*312 tons Vessel = 9*575 tons Total in hall = 23,000 tons Jo Lister, CRPP-EPFL, ITER Data Challenges, ICALEPCS 2005, Geneva, October 2005

  4. ITER’s data rates and volumes - OK • Conclusion in 2003 • Rate is probably not a problem with existing technology • Volume is not a problem • But it will be hard work of course, even if we do have solutions • … but…. we can not obviously USE the data we are archiving • This will be a challenge, mentioned by Martin Greenwald • Knowledge or model-based filters will have to be developed • That was the feeling 2 years ago • Take the potential source rate 100 GB/s peak • Take 500’000 to 1’000’000 hardware channels, 0.1 – 108 Hz • Take 1PB per year Jo Lister, CRPP-EPFL, ITER Data Challenges, ICALEPCS 2005, Geneva, October 2005

  5. So what are we thinking about for CODAC ? • ITER COntrol Data Acquisition and Communication architecture will consider • Political “in kind” procurement • Procurement interface definitions • Procurement standards definitions • General life-cycle issues of procured systems, associated perenity issues • ITER data flow performance : data rates and volumes - technical • Combining acquisition and control into “data flow” - tokamak specific • Networking inside and outside an ITER sanitised area - topic this week • Remote participation and remote operation - covered by Martin Greenwald • Remote maintenance of procured systems • Discharge design and modifications - tokamak specific • Discharge monitoring and control - tokamak specific Jo Lister, CRPP-EPFL, ITER Data Challenges, ICALEPCS 2005, Geneva, October 2005

  6. Procurement, integration and operation • Most of procurement of “External” systems will be “political” in-kind • Procurement of the “internal” systems (IT infrastructure) is central • Parties (e.g. EU, USA) will have to deliver full systems with an agreed value • They will procure through their own administration • Procurement may involve research institutions • Manufacturing may involve local/foreign industry • Industry may outsource some of the components • Then it will be integrated under high visibility and operated as fast as possible • How do we optimise this integration ? Jo Lister, CRPP-EPFL, ITER Data Challenges, ICALEPCS 2005, Geneva, October 2005

  7. Major issues • Can we aim at plug-and-play of all “external systems” ? • Standards for hardware • Standards for software • Standards for data exchange • Security • Respect for these standards is implicit, but is it guaranteed ? • Google “plug and play” is interesting, the top hit is…(from Microsoft) This update resolves a newly-discovered, privately-reported vulnerability. A remote code execution vulnerability exists in Plug and Play (PnP) that could ... • Are we being naïve - if so, we should stop now ! • Are we being visionary - if so, we should be brave ! Jo Lister, CRPP-EPFL, ITER Data Challenges, ICALEPCS 2005, Geneva, October 2005

  8. Interlock Network 2 Network 1 ITER Specs System Local control Local store Water/air/power/hydraulics What do we mean by a system ? Jo Lister, CRPP-EPFL, ITER Data Challenges, ICALEPCS 2005, Geneva, October 2005

  9. Hardware standards • Can we impose hardware standards on what is inside a procured system ? • Is it better to have a supplier work with what he knows best ? • Is it better to have common spares ? • Who is responsible for maintaining the hardware after 10-20 years ? • Who will actually end up keeping old equipment going ? • Remember we are talking of large planetary differences in work practice and technology Jo Lister, CRPP-EPFL, ITER Data Challenges, ICALEPCS 2005, Geneva, October 2005

  10. Software standards • Can we impose software standards on what is inside a procured system ? • Is it better to have a supplier work with what he knows best ? • Is it better to have common products ? • Is it necessary to have common methodology ? • Who is responsible for maintaining the software after 10-20 years ? • Who is responsible for software updates (security?) for 20 years ? • Who will actually end up keeping old software going ? • Remember we are talking of large planetary differences in work practice and technology Jo Lister, CRPP-EPFL, ITER Data Challenges, ICALEPCS 2005, Geneva, October 2005

  11. Data exchange standards • Can we impose data exchange standards on a procured system ? • Is it better to have a supplier work with what he knows best ? • The other software issues are all here again • Remember we are talking of large planetary differences in work practice and technology • Can we use internationally recognised (i.e. not fusion-specific) norms ? • Do we have to copy others and make an ITER-specific generic framework ? Jo Lister, CRPP-EPFL, ITER Data Challenges, ICALEPCS 2005, Geneva, October 2005

  12. Remote integrators for procurement • Can ITER procure at the level of wiring ? – ITER will never enough staff • Can ITER procure at the level of subsystem ? - homogeneity, staff • Can ITER procure at the level of full “identical” systems?- norms • ITER-qualified/trained local integrators could be a good solution – multi-national Jo Lister, CRPP-EPFL, ITER Data Challenges, ICALEPCS 2005, Geneva, October 2005

  13. state state state External External External External Internal Internal Internal Service communication RTDB1 RTDB2 Service Oriented Architecture • Could ITER go this way ? • This model is natural when designing an integrated plant, when integration is not just the last act • It is • extendable • not necessarily scalable • has hidden middleware • has a hidden framework • has a general API • Each system can interface to a given API e.g. VSYSTEM, PVSS, EPICS ?? Jo Lister, CRPP-EPFL, ITER Data Challenges, ICALEPCS 2005, Geneva, October 2005

  14. Integration of complex equipment • “Slow controls” will generate a a lot of data, but ITER will have slower time-scales than present tokamaks. We assume 500’000 to 1’000’000 channels • This is data handling rather than technical - we have to use and understand this data • Requires more effort on Finite State Machine system description than on hardware/software techniques • Corresponds again to a structured data view of the world • Needs a universal description of the FSM- see ongoing work elsewhere • Generating the working control software will still require lots of inventiveness, but most of all will require a complete data-driven description of the systems • Can we combine plug and play devices into a full inter-operating system? Jo Lister, CRPP-EPFL, ITER Data Challenges, ICALEPCS 2005, Geneva, October 2005

  15. Security • We wish remote users to be able to service their equipment • This is inevitable • However, this opens the door into the machine control from outside • We can impose a gateway, content sensitive, to relay instructions coming in, with strict control (like PVSSII), but will the hackers win? • Can we guarantee that the outside world cannot penetrate a content-sensitive gateway in one direction ? - in 8 years time…. • The other security issues are conventional - i.e. extremely difficult Jo Lister, CRPP-EPFL, ITER Data Challenges, ICALEPCS 2005, Geneva, October 2005

  16. Vision of ITER integration in a SOA • SOA are in and out of favour, but were “sold” for Enterprise Application Integration in heterogeneous environments with dominant legacy problems. • Is it the right model for integrating complex physics plant ? • What Web Services appear to offer is • Strong control over the communication method between supplier and ITER • Strong content over the data exchange content • Internationally established norms, internationalisation • Security, traceability • Management transactions for configuration, commands, documentation, description, essential for plug-and-play • Inadequate performance for the sustained high rate data flow- probably Jo Lister, CRPP-EPFL, ITER Data Challenges, ICALEPCS 2005, Geneva, October 2005

  17. Summary • ITER will presents a tremendous scientific and technological challenge for the optimal handling of all the data of the project • Procurement will bring interesting international challenges • Architecture must consider the procurement and integration problems • “Strong control” does not create a partnership, the SOA is an enabling technology, not a guarantee • How do we minimise the risks of less than total success ? • On time • On cost • Required performance Jo Lister, CRPP-EPFL, ITER Data Challenges, ICALEPCS 2005, Geneva, October 2005

  18. Round Table • We are opening the round table discussion on procurement for large international research projects • We hope that you will stay and share your advice and experience Jo Lister, CRPP-EPFL, ITER Data Challenges, ICALEPCS 2005, Geneva, October 2005

More Related