E3DDS SG Meeting 20190123

From neicext
Jump to navigation Jump to search


E3DDS steering group meeting

Time: 2019-01-23 10h00-12h00 CET

Place: https://cscfi.zoom.us/j/9917669499

Invitations:

  • Craig Heinselman, EISCAT
  • Yasunobu Ogawa, EISCAT
  • Tomasz Malkiewicz, NeIC, Project owner
  • John White, NeIC, Project manager

Agenda

Presence, approval of agenda (5')

Project status

Project plan

The project plan is being followed:

- PDF form at: https://wiki.neic.no/wiki/EISCAT_3D_Data_Solutions#Documents - The google docs version is at: https://docs.google.com/document/d/1rgO0qtHJR6zjigdQ2KytFD3ySLH11XvasvSpJNCBVGg/

  • The Delivery 1 document was planned for October 2018... sent to the SG on Dec 17th 2018 (just over a month behind)
  • Project works well... good attendance to meetings

https://wiki.neic.no/int/Category:E3DDS_Project_minutes

  • Deliverable documents are "owned" by the partners. So far has worked well.

Deliverable 1

  • As document written, understand the missing parts and information.
  • Second beam former... "chicken and egg" for the number and type of nodes (expensive) required.
  • Summation of beam results... in-server or in-network.
  • GPU (and FPGA) being investigated.
  • Differences between the TX and RX sites.
  • Minimum requirement is to process low bandwidth (5 MHz) data at high-power setting in real time.
  • Wide-bandwidth operations "sporadic".
  • Web-page of input network and computing options.
  • These summarize the various input network and SBF computing options considered so far... See:

https://wiki.neic.no/int/E3DDS_Deliverable_1#Input_network.2FSBF.2FRB_nodes_options

Deliverable 2

  • Working on the Deliverable 2: The on-site data processing simulation with a baseline of a prototype second beamformer simulation is written and supported to run in the national provider e-infrastructures.
  • Second beam former tested on Kebnekaise in Umea, AMD servers in Slovenia (NeIC NT1 contact and soon to be GPU-equipped IBM Power9 cluster in NY (also NeIC NT1 contact)
  • Gold-plated case for testing the system: tracking rocket launch.
  • Question to SG: How to simulate the system for space debris? Johan K and Daniel K?
  • Simulated test of the radar control system?
  • Should this be taken as a simulation case?

Milestone 1 CodeRefinery workshop

  • In conjunction with IRF
  • Meeting with EISCAT_3D project leader, new staff.
  • Useful... get priorities.
  • FPGA discussions (got some contacts at PDC and Linkoping)

Milestone 2

  • Dialog with software architect
  • Overall software architect still TDB?
  • We meet with Harri, Assar, Simon weekly

New Topic?

  • Question from Johan/Henrik to assess the cluster management needs for EISCAT_3D.
  • Discussed in PG meeting 2019/01/21
  • One issue to resolve... who can run the clusters? EISCAT-employed or EISCAT-controlled personnel (out sourced).
  • Check the requirements for secure computing e.g. plenty of experience within NPs handling patient data.
  • The issue of white/black listed orbits still not clear. Removed in FBF and/or prompt computing? Probably more likely at the FileWriter. Current EISCAT cannot give orbits... 3D can.
  • Related to Operational Profiling... e.g. Space Debris is a good example.
  • Appendix or section in the NT1 Deliverable.

Operational Profiling

  • On going, not so slow now...
  • But, simulation cases (see above) help now.
  • Another case, e.g. the low power 5 MHz default operations?

Stakeholder meeting

  • Would like to invite Johan K or Daniel K to present the Space Debris?

Documents from E3DS

  • MC-3
  • Tried to follow up on two contacts from the last Stakeholder meeting. BenchIT (NO) good work done there in making clear what it costs to produce IT in an academic environment. Can BenchIT can help EISCAT-3D?
  • Ongoing...

E3DDS future

  • Follow-up or transferral of results
  • Preparation of proposal to for the NeIC Open Call.
  • Pre-announcement of the open call in January. Expected deadline May.

Next meeting

Proposal: (Early) May 2019 F2F at Uppsala, 14 -16 CEST (conjunction with the EISCAT User meeting).

AOB

  • Sites decision requested by National Providers

Minutes

Presence, approval of agenda (5')

Present: John, Tomasz, Craig, Yasunobu

The agenda is approved.

Project status

Project status given by John.

  • The status of the project looks fine.

Project plan

Deliverable 1

  • The document is accepted by the SG. Can be moved to the public page (action).
  • The document is to be updated when most useful for the EISCAT_3D project. To be coordinated with Johan and Henrik (also project group). Action.
  • There are these Appendices A and B. Are these future outlooks still valid? (Yes).
  • How did you decide on the minimum requirement for the data chain of 5 MHz full power? (Used the EISCAT_3D project minimum. From the technical description)

Deliverable 2

  • Invite Johan K/Daniel K to the Stakeholder meeting for Space Debris
  • Also invite Ingemar or Assar for rocketry. (Action)

Milestone 1 CodeRefinery workshop

  • Please note.... the CodeRefinery will not run for ever. The repositories will need to be taken over!

Milestone 2

  • Noted that the discussions with the software people are ongoing each week.

New Topic?

  • Decision: This is a useful topic to follow. Consult Johan/Henrik for what and when.
  • John to email with Johan and Henrik to see how to include the cluster management work. (action)

Operational Profiling

  • Noted. No decisions made.

Stakeholder meeting

  • Craig should be able to attend. Yasunobu not.
  • Action: John to send reminder email. Start building the agenda.

Documents from E3DS

  • Try sending another reminder to Norway about the BenchIT. Also follow up on the EuroHPC work. Action.

E3DDS future

  • The deadline will be in late May 23rd or 30th. We need to decide on whether to do in May 2019 or 2020?
  • Need to determine why it's bad for this cooperation to end in April 2020... what's missing?
  • Case ready for Stakeholder meeting? March 27th?
  • John to coordinate the case. (Action)
  • We agree on a meeting 09h00 March 27th, Espoo (16h00 Japan)

Next meeting

  • See above... March 27th 09h00-10h00 Espoo (Yasunobu by remote)

AOB

  • The "sites" decision... More information?
  • As for the processing sites there is no progress. We are not focussing on this right now. Delivery of antennas. Permits for land development etc. There are also some additional tenders. We need to install e-infrastructure that we can afford. But not over-commit. We are not going to have three data centres.
  • Are there any updates on timetable?
  • We are still on schedule... operational by the end of 2021. One year of cushion... but start to push against this schedule. This is operations for customers... i.e. scientists in each member nations.
Retrieved from "https://wiki.neic.no/w/ext/index.php?title=E3DDS_SG_Meeting_20190123&oldid=10388"