JIVE Wiki

The kickoff telecon was held on 2010-08-27 and started at 12:00 CEST.

Attendees: Frank Blankman (Nordunet), Paul Boven (JIVE), Paul Burgess (Jodrell Bank), Peter Hinrich (Surfnet), Richard Hughes-Jones (Dante), Neal Jackson (Manchester Uni), Brian Bach Mortensen (Nordunet), Chris Phillips (CSIRO), Ralph Spencer (Manchester Uni), Arpad Szomoru (JIVE), Tasso Tzioumus (CSIRO), Charles Yun (JIVE).

Agenda:

1.) Introductions

2.) Overall planning (deliverables, manpower, hiring, hardware)

3.) Short term planning (upcoming milestones and deliverables, esp. MS601)

4.) Consortium Agreement

5.) Upcoming NEXPReS Board meeting

6.) AoB

7.) Scheduling follow-up telecon/meeting

Minutes:

2.) Overall planning

Task 1:

Peter Hinrich (Surfnet): Gerben van Malenstijn and Hans Trompert from SURFnet have been assigned to work on the NEXPReS projects, they are also active in the Gigaport3 implementation.
Paul Boven (JIVE): We have just published a job opening at JIVE and have already received our first replies.
Frank Blankman (Nordunet): Nordunet will contribute mainly in the technical implementation of this task.
Tasso Tzioumus (CSIRO): Tasso, Chris Phillips and Shaun Amy will contribute to this task. Aarnet does not currently have BoD services available but are keenly interested in the progress in this field.

Task 2: This was skipped as nobody from ASTRON could make it to the telecon.

Task 3:

Ralph Spencer (Manchester Uni): We have drafted an advertisement for a position that is currently being prepared for publication. We will build on our previous experience with both UDPmon (together with Richard Hughes-Jones) and the iBob network tester. There might already be iBobs available to start working on.
Richard Hughes-Jones (Dante): There is 10G FPGA hardware and software available for testing. The software is modular and should be easy to integrate with the NEXPReS controls. We will begin with testing the equipment that's available in Manchester and get it to work again.

Task 4:

Peter Hinrich (SURFnet): SURFnet is making preparations to enable sending multiple-Gb/s streams to JIVE from Amsterdam, and is currently investigating which equipment and other resources would be needed for that. Also changing DRAC to work for paths that have more than 1Gb/s of bandwidth.
Brian Bach Mortenson (Nordunet): What parameters would these paths have, in terms of packet loss, latency and Jitter?
Paul Boven (JIVE): Delay and even jitter are not very big issues for the e-VLBI application. We expect that the higher speeds will be generated by FPGA based equipment at the telescopes, and will therefore be nicely paced.
Richard Hughes Jones (Dante) reminds us of previous experiments he conducted where bursty traffic caused problems when trying to send data through a 4Gb/s path with 10Gb/s based equipment.
Frank Blankman (Nordunet): Nordunet uses a dynamic tool to set up MPLS based paths between Oscars. Internally paths can be from 1 to 40 Gb/s, with up to 10Gb/s for customer ports. In Glif there have already been interdomain tests between Surfnet and Nordunet.
Richard Hughes Jones (Dante): In Dice (?) we are also working on interactions between Geant and Oscars in the ESNET in the USA.
Brian Bach Mortenson (Nordunet): In Geant 3, we are working on a service description for the On-demand Bandwidth service. We hope to have a first version available at 5 large NRENs within half a year, though they will need to integrate this with their networks, too. We will not offer all the functionality at the start yet.
Paul Boven (JIVE): will send Brian a list of countries/telescopes for both e-VLBI and LOFAR so we can see how well it matches the first stages of the GN3 BoD service.

3.) Short term planning (upcoming milestones and deliverables, esp MS601)

Paul Boven (JIVE): Our first upcoming milestone will be MS601 in month 4 (end of october) of this project, as listed in the WP6 description under Task 1. We need to get started quickly on particularly this milestone to have the report ready in time.
Richard Hughes-Jones (Dante): The Network Service Interface WG of the OGF has put up a draft document describing some of their work, which he'll point us to. Geant3 SA2, JRA2 and NA4 are involved in that work. Brian Bach Mortensen: Nordunet can help with working out the architecture and inter-domain issues. There will be an Autobahn workshop later this year.
Peter Hinrich (Surfnet): Surfnet should be able to help with drafting this report.

4.) Consortium Agreement

Paul Boven (JIVE): All partners have received the consortium agreement: please make sure that your institute is looking at these, and providing feedback or even better a signed version.
Charles Yun (JIVE): We can only start disbursing funds to your institute once you have returned a signed copy of the consortium agreement.

5.) Upcoming NEXPReS Board meeting:

There will be a NEXPReS Board meeting on 20th of september in Manchester, prior to the EVN symposium.

Ralph Spencer (Manchester): All arangements have been made, everything is booked and ready.
Paul Boven (JIVE): all workpackages, including this, will make a presentation to the NEXPReS board. A draft version will be circulated for comments.

6.) Any other Business:

Paul Burgess (Jodrell Bank): The current iBob boards have limited memory capacity and we might want to use their folluw-up design, the Roach board, for these tests.
Richard Hughes-Jones (Dante): We should make sure we don't exclude the possibility of using IPv6 when necessary, although we'll most likely only need IPv4.
Frank Blankman (Nordunet): why is IP needed for these paths? Paul Boven (JIVE): the paths themselves might be just ethernet based, but it is convenient for getting from telescope to POP, and for the internal routing at JIVE, to use IP addressing for the traffic. It also makes testing/debugging a path much easier.

7.) Scheduling of the next telecon/meeting:

Paul Boven (JIVE): we propose to hold another telecon in the first weeks of October, to coordinate work on our first report.
Richard Hughes-Jones (Dante): Should we have a meeting in Manchester during the NEXPReS board meeting?
Paul Boven (JIVE): we planned this telecon earlier to be able to not delay work on this Work Package too long. It seems that quite a few participants will be going to Manchester, so together with Ralph Spencer we'll work out if this is possible.