LHCOPN-LHCONE meeting - Amsterdam

Other Institutes

Other Institutes

Science Park - Amsterdam NL

Venue: New SURFsara building, at Science Park 140 (just 100 meters north east of the previous SARA building at S.P. 121), room SURF-meetingroom (ground floor)

Travel directions: Public Transport

Hotels suggestions:
Reachable by bus 40/240:
- Hotel Casa 400

Reachable by tram 9 (15 minute walk from SARA or take bus 40/240):
- Eden Lancaster Hotel
- Eden Amsterdam Manor Hotel
- Tropen Hotel

Other hotels near Amsterdam Central Station (city center) are easily reachable by train from the Amsterdam Science Park station (5 minute walk from SARA).

  • Andrew Lee
  • Bas Kreukniet
  • Bill Johnston
  • Bruno Hoeft
  • Christos Papadopoulos
  • Dale M. Finkelson
  • David Groep
  • David Mitchell
  • David Salmon
  • Edoardo Martelli
  • Erik Ruiter
  • Erik-Jan Bos
  • Fernando Lopez
  • Freek Dijkstra
  • Guido Aben
  • Hsin-Yen Chen
  • Jerome Bernier
  • Joe Metzger
  • Marco Marletta
  • Mian Usman
  • Michael O'Connor
  • Patrick Dorn
  • Peter Clarke
  • Philip DeMar
  • Ramiro Voicu
  • Ronald van der Pol
  • Sander Boele
  • Shawn Mc Kee
  • Silvio Pardi
  • Tony Cass
  • Tristan Suerink
  • Vincenzo Capone
  • Yatish Kumar

A brief update on the status of the LHCOPN was given. The OPN has been quite stable recently  and the only changes concerned some link upgrade and IPv6 activations. The traffic has increased with the start of Run2 and few sites are considering upgrading the capacity of their links to CERN.

CERN has announced a change of its LHCOPN and LHCONE prefixes. The change aims to reduce the number of servers exposed. Exact dates will be announced to the mailing lists.

The discussion about stop using LHCOPN for T1-T1 traffic got more input.
- NL-T1 was against, because LHCOPN is cheaper than LHCONE.
- GEANT supports the current status, i.e. Tier1s are allowed to decide bilaterally the transit policies
- UK-T1-RAL stated that the UK sites don't need LHCONE for the time being. If T1-T1 traffic has to be moved to LHCONE, then there must be a compelling reason to do it. If it was the case, the Experiments should be involved in the decision making.
- CERN is happy to keep providing transit to T1-T1 traffic. T1s must be aware that most of the backup paths have already been moved to LHCONE, though.
In the end has been agreed to not force the removal of T1-T1 traffic from LHCOPN. And to let Tier1s decide with their peers where to put that  traffic (symmetry should be respected).


L3VPN update: Mian Usman showed traffic statistics for the main VRF providers. He showed also the interesting information which can be provided by the flow analyser tool that Geant has adopted.

L3VPN operations update: Mike O'Connor presented an analyses on the utilization of BGP communities in LHCONE and the new tools to check the content of the ESnet routing table.

LHCONE AUP audit: the AUP has been approved by ~70% of the sites. Edoardo Martelli will contact the missing sites to get the acknowledgements. Pierre Auger has not provided the necessary information, so it has not been formally added to the AUP.
It followed a discussion about connecting commercial cloud providers to LHCONE. The Network providers already peer with cloud providers, which routes are not sent to  peering partners. They expressed some concern in transiting commercial traffic over R&E networks.

perfSONAR update: Shawn McKee presented the progresses made and new tools using ElasticSearch and MadDash

Report on Asian Tier Centre Forum in Korea: Bill Johnston gave an extensive report on the forum held in Korea which saw the participations of most of the WLCG Asian sites and important progress in realizing VRFs in that region.

Transit VRF: more LHCONE VRFs are being implemented and it's not always possible to guarantee a full mesh among them, especially for those outside Europe and North America. For them it's important to find transit VRFs that can assure global reachability.
Currently the default routing policy is Peering, so ad-hoc agreements have to be arranged in order to grant VRF transit.


BGP-based solutions with dynamic circuits: Magnus Bergroth presented possible solutions using BGP to the L3 routing issue of sites using p2p circuits. A solution is based on Route servers interacting with BFD for knowing the status of the data plane.

Named Data Networking: Christos Papadopoulos presented NDN, a project that aims to implement data distribution using data identifiers rather than host addresses.

Progress of SDN, OVS and dynamic circuits: Ramiro Vocu presented the progresses made with integrating NSI circuits with PhEDEx using OVS (Open vSwitch) and how they were able to shape the bandwidth on a p2p circuit. He also showed the result of data transfers using servers with 100G NICs.

BoD service plans in GÉANT: Mian Usman presented how BoD is currently implemented in Geant and it will evolve using the DynPaC framework, ONOS and SDN

Automated GOLE update: Gerben van Malenstein showed the activity of the AutoGOLE project, the dashboard they have implemented

*Next meeting*

The next LHCOPN and LHCONE meeting will be hosted by the ISGC Symposium in Taipei on the 13-14 of March 2016:

There are minutes attached to this event. Show them.