Report from Task1

• Simon Casey

Simon: Investigated CPU goveror effect but this seems to have no effect on throughput. A new system has been set up but sending UDP flows to the old fast system see 20% packet loss, sending to the old slow system only see fractional packet loss. libvma UDP tests from old slow host to old fast host: packet spacing us throughput loss 0.9 78 none 0.8 87 5% loss Richard: Further UDP flow tests between the GÉANT DNT nodes in the lab with the Connect X5 NIC showed packet loss caused by the NIC not having sufficient buffer space for the incoming packets. increasing the size of the RX ring buffer to 4096 or 8192 fixed it. Simon has also seen this. RDMA test program being put together. Oleg: He has sent round some links about the xrootd protocol: The the brief background about xrootd you can find in the slides G. DONVITO : http://www.desy.de/dvsem/WS1112/donvito_talk.pdf and xrootd redirection well explained : https://twiki.cern.ch/twiki/bin/view/Main/RedirectorsSubscription There is a 10Gigabit testbed available at Julich that could be used for the WP4 tests. Action Richard & Simon: exchange ssh Keys and system login info.

Report from Task 2

• Mauro Nanni

Mauro sent round his ideas on collecting general prices from the NRENs. Action Richard: Check on the best approach with GEANT BusDev team.

Report from Task 3

• Richard Hughes-Jones (GEANT)

Richard: Rosie Bolton (SDP Cam and SKAO) has posted some documents from SDP, see https://drive.google.com/drive/folders/0B8aDq0j3MqAQRF9CRldGWFhNMTA

Report from Task 4

• Richard Hughes-Jones (GEANT)

Mary: ASTRON is planning to upgrade their site infrastructure with a new Juniper boarder router; it will have 10 gigabit links but no 100 gigabit yet. the access link is expected to remain as 4 * 10 GE. SURFnet will be talking with JIVE to determine their 5 year plans. A list of European sites that are 100 Gigabit capable would be helpful. Action Mary & Richard: Start the list of 100 Gigabit capable sites. Shaun: The AARNet 100 Gigabit test system is at the MRO, Tim Rayner is not back from vacation. He has 2 Mellanox ConnextX5 cards and will chase the new test server for MRO. For the Enlighten your Research flows from MRO to ASTRON, the Tile Cards (TPM) are connected to a Mellenox 40 gigabit switch that has spare ports, so it should be ok to send a 10 gigabit raw voltage stream to ASTRON for corellation. Expect tests in October.

Planning for the AENEAS Face to Face meeting 18-20 October 17

• All

Date of next Telcon 21 August 2017

AOB 1. It would be useful for WP3 and WP4 to have a telcon prior to the AENEAS F2F meeting to see how they fit together. Action Richard: Discuss with Anna to find a date. See: http://doodle.com/poll/utdydhpczhmvbfqe 2. Shaun: is there any information on data placement for SKA regional centers? a tiered model might work for HEP but geographically located data placement might be better for Astronomy. [RHJ: Note after the SKA Network group meeting. SKAO has approached the science working groups, but ideas are still being discussed. Chicken and egg - where are the data centres / where to put the data.]