LHCONE R&D and RNT-WG call

Tuesday 20 Jun 2023, 17:00 → 18:00 Europe/Zurich

Edoardo Martelli (CERN), Joe Mambretti (International Center for Advanced Internet Research Northwestern University), Marian Babik (CERN), Shawn Mc Kee (University of Michigan (US))

Presents: Edoardo martelli, Marcos Schwarz, James Walder, Federica Agostini, Katy Ellis , Doug Southworth, Andy Lake, William Johnston, Garhan Attenbury, Tristan Sullivan, Julio Ibarra, Carmen Misa, Li Chen, Michel Lamber, Tristan Sullivan, Phil DeMar, Shawn Mckee, Asif Shan, Joe Mambretti , Laurent Duflot. Bruno Hoeft, Marian Babik, Xiaomei Zhang, Tim Chown. Duncan Rand

SC23
- NRE for SC23 demos of NOTED and packet marking were sent by Joe. They have been accepted, confirmations have been received. Note that the NRE  submission are drafts and  can still be modified . The deadline for the final version is in mid-January .
NOTED demo:
- the intention is to build on what demonstrated last year (SENSE links to KIT, TRIUMF and CERN) with the addition of links to Tier2s and to FNAL. Possible Tier2s: UCSD, Caltech (already connected to SENSE) and AGLT2. 
- FNAL is implementing testbed and will have a 100G link to that testbed. Asif Shan is a new person working at FNAL who is dedicated to developing the testbed
- AGTL2 is already very busy and may not have time for a NOTED demo. 
- Joe and Edoardo will contact Caltech and UCSD.
Packet tagging demo:
- On the list  is University of Nebraska (XRoot), KIT, University of Victoria, CERN, hopefully AGLT2 (using dCache)
- Collectors: ESnet global collector, SFlow collector at ESnet/SCinet, CERN P4 collector
- There is a regional collector in the UK, but no UK sites are participating. 
- Tristan would like a 400G from Seattle to Denver and a 400G DTN on the SC23 venue show floor. Joe says there should be both. (PacWave Seattle 400GóStarLight<1.2 TbóSC23)

Data Challenge 24
- Anything more we want to test?
- Shawn gave a presentations to ATLAS and CMS about it. The plan was presented backward, from what they want to show during  DC24 to the mini challenges that we want to run in their preparation.
https://docs.google.com/presentation/d/1j6yrQv6q2w-BZlsQp9EKdUt8mjbCfBUECvMJ2nAWmVM/edit#slide=id.p1
- The list of Tier2 that will participate is being prepared by DOMA
- The target is 25% of the HL-LHC  target. DOMA is making a plan that will make it possible to estimate the impact on the network. The challenge of this task is to estimate the impact on/from the Tier2s, which were not included in the previous DC (Tier2 connectivity is more complex than T1s, especially if full mesh connectivty is expected)
- Duncan suggests that every site run individual tests to check what their able to achieve, then to sum up all the results to have a rough idea of the total load.
- All Tier1s will participate

- There was a meeting of WLCG monitoring that stressed the importance of having the tagging to understand the VO generating the traffic.
- Tim asked if there are dashboards available for flow labelling/fireflies. Andy: ESnet has already some simple dashboards (but they also show sflow data ESnet is collecting).
- Tristan: Would it be worth to try to mark the traffic on the storage that is dedicated to only a VO? For storages running xroot and flowd this is already possible and relatively easy to setup. 
- It remains to be seen if NRENs would be able to perform accounting on the markings (IPv6 flow labels) as this requires HW equipment. SC23 flow label visualisation with sFlow is documented at https://blog.sflow.com/2022/11/scientific-network-tags-scitags.html - according to the blog any sFlow capable device should work, full list is at https://sflow.org/products/network.php )

Actions for DC24:
- Understand from DOMA which Tier2s will participate in DC24 and understand the impact on their links and if they're willing to participate in the packet marking.
- Think of more challenges that could be shown

RNTWG

Packet Marking
- Started testing dCache at AGL2. There are some issues with start fireflies that we need to debug.
- Python Flowd EPEL: no news.
- Started discussion with EOS and FTS at CERN about tagging implementation details. Agreed to work on a concrete proposal with details. Developers have already started looking at the Scitags specs.
- Discussing proposal to simplify HTTP/TPC specification (way to propagate flow IDs from data management to the storages)
- It was proposed that we organise a dedicated meeting to discuss packet marking SC23 and DC24 preparations (have more in-depth discussion on collectors, scalability, visualisations, etc.)

Packet Pacing
- First meeting took place (notes can be found at https://indico.cern.ch/event/1287745/ )
 

AOB
- Suggested to have an Indico page for these calls.
- Shawn: people in ATLAS start being worried of the amount of work needed for these challenges


Next meeting
- Next meeting exceptionally on  Wed 5th of July (Tue 4th is holidays in the US)

17:00 → 17:30 Preparation of SC23 demos and DC24 mini-challenges

17:30 → 17:45 RNTWG

17:45 → 17:55 AOB

17:55 → 18:00 Next meeting