LHCOPN meeting - London

Europe/London
Woburn House Conference Centre

Woburn House Conference Centre

20 Tavistock Square London UK
Description
Host: JANET(UK)
Venue: Woburn House Conference Centre London, UK
Accomodation: Ask Martin Lewis at Destination Solutions (free of charge service)
Participants
  • Alberto Escolano Sanchez
  • Artur Barczyk
  • Aurelie Reymund
  • Bruno Hoeft
  • David Foster
  • David Salmon
  • David Salmon
  • Edoardo Martelli
  • Emma Apted
  • Eric Boyd
  • Gerard Bernabeu
  • Guillaume Cessieux
  • Hanno Pet
  • Joe Metzger
  • John Shade
  • Jurriaan Saathof
  • Lars Fischer
  • Magnus Bergroth
  • Nick Moore
  • Otto Kreiter
  • Phil DeMar
  • Ramiro Voicu
  • Robin Tasker
  • Stefano Zani
  • Szymon Trocha
  • Warren Matthews
  • William Johnston
    • 14:00 14:01
      Meeting Start 1m
    • 14:01 14:10
      Welcome and Introduction 9m
      Slides
    • 14:10 14:20
      LHCOPN News 10m
      Speaker: Edoardo Martelli
      Slides
    • 14:20 14:50
      UK-T1-RAL update 30m
      Speaker: Robin Tasker
      Slides
    • 14:50 15:20
      JANET(UK) update 30m
      Speaker: David Salmon
      Slides
    • 15:30 15:50
      Coffee break 20m
    • 15:50 16:40
      LHCOPN network monitoring at the Tier0/1 sites 50m
      • UK-T1-RAL 5m
        Speaker: Robin Tasker
        Slides
      • DE-KIT 5m
        Speaker: Bruno Hoeft
        Slides
      • NL-T1 5m
        Speaker: Hanno Pet
        Slides
      • IT-INFN-CNAF 5m
        Speaker: Stefano Zani
        Slides
      • CH-CERN 5m
        Speaker: Edoardo Martelli
        Slides
      • NDGF 5m
        Speaker: Magnus Bergroth
        Slides
      • FR-CCIN2P3 5m
        Speaker: Guillaume Cessieux
        Slides
      • ES-PIC 5m
        Speaker: Gerard Bernabeu
        Slides
      • US-FNAL-CMS 5m
        Speaker: Phil DeMar
        Slides
    • 17:00 17:30
      Monitoring Working Group outcomes 30m
      Speaker: John Shade
      Slides
    • 17:30 18:00
      Discussion: LHCOPN monitoring 30m
      - current use of the MDM by the Tier0/1s - requirments for the LHCOPN monitoring system
    • 18:00 18:01
      Meeting end for the day 1m
    • 19:30 20:30
  • Tuesday, 9 March
    • 09:00 09:01
      Meeting re-start 1m
    • 09:01 09:25
      LHCOPN Operations update 24m
      Speaker: Guillaume Cessieux
      Slides
    • 09:25 09:30
      Transition of the ENOC activities post EGEE 5m
      Speaker: Guillaume Cessieux
      Slides
    • 09:30 09:50
      Internet2 activities for the US LHC community 20m
      Speaker: Eric Boyd
      Slides
    • 09:50 10:10
      Monalisa capabilites for the LHCOPN 20m
      Speaker: Ramiro Voicu
      Slides
    • 10:10 10:30
      coffee break 20m
    • 10:30 11:00
      Tier2s networking 30m
      • European Tier2s to US Tier1s traffic analysis 20m
        Speaker: Bill Johnston
        Paper
    • 11:00 11:40
    • 11:40 11:50
      Meeting wrap-up and conclusion 10m
      News: BNL and FNAL primary links now 2x8.5Gbps RAL 10G backup link will be ready in 2/3 months. PIC backup link now on a different path inside the Geant network. RAL update New Computer centre building. Some issue, now sorted out. The second 10G link for LHCOPN will be delivered in 2/3 months. Currently design the next generation network architecture. JANET update Current network: SuperJANET5 until 2013. Services: JANET IP (40G core), JANET Lightpath (Ethernet over MPLS), JANET Aurora (dark fibre research platform) Positive 100G trials with Nortel and Ciena Monitoring RAL: using the Netsight of Janet, Cacti at the site (weathermap), Scrutinizer (commercial, sflow netflow) KIT: ping, nagios, Netview (sends email and sms), Cacti (BGP monitoring), Netflow (Cisco product), Syslog NL-T1: Nagios, Cacti, PHP-Syslog-NG, Rancid, Remedy. BGP monitored by Nagios. Availability: ping from nagios. CNAF: MRTG(traffic) NAgios (status availability and alarms), Netflow analyzer Pro, CERN: spectrum (commercial, link status and stats), Nagios (BGP NDGF: Nagios, ZINO (home maded, link status, BGP), sierra (home made sflow collector) Alcatel and Ciena NMS. IN2P3: Cricket for statistics, smokeping (remote hosts), Netsurv (home made) PIC: Cacti, Nagios (BGP, interfaces) FNAL: Nagios, Cisco IP SLA, RoutePlotter (traceroute and RTT changes monitoring) Netflow analysys. ESnet: Spectrum (devices, links, LSPs. Main tools for the NOC). E2EMON (but not used becaouse slower in reporting) USLHCnet: Monalisa (weathermap, link status and availability, topology chnages, send alarms), Perfsonar-PS (feeds E2emon) Monitoring WG Who is using perfsonar MDM: CNAF is using the MDM to check the network is able to run 1G data transfer between T1-T1. KIT: No, only checking it displays correct information NL-T1: No CNAF: yes: data transfers, hades, weathermap, OWD, traceroute CERN: no NDGF: no In2P3: no PIC: no, especially because the backup link cannot be added. Would like to have jumbo support. FNAL:no Comments: Should be only a measurement system, not alarming/monitoring Who are the users? Application managers, network engineers. Operations Twiki not updated Integrate or tickets in the the standard GGUS system? Pro and cons. No email interface to ggus will be implemented. Support may be reduce after the end of EGEE. We need a SLD and monitoring. ENOC after EGEE EGEE ends April 2010, EGI should follow. Only 0.5 FTE for networking is foreseen. LHCOPN operations support will be followed by IN2P3 GGUS TTS support, included LHCOPN part, is moved to WLCG Internet2 Installing perfsonar-ps in many US T2s and T3s. Monalisa Large installed base. Serving Alice. ESnet on US tier2s A increase of traffic between US T2s and European T1s is foreseen. Dedicated links would be better than using the generic IP connectivity as today. Monitoring Discussion NL-T1: access to raw data not very useful. status map accessible to all users via grid certificate. KIT: BGP route checks for change of paths, data transfers. PIC: E2Emon would be useful. Perfsonar is useful to show the link is not the problem. Not interested in row data. Weather map access should be open. NDGF: Wish access to raw data. No use of e2emon FNAL: have measurements, notifications, weathermap for end users, should be service based monitoring and not link based CNAF: it's useful, access to raw data may be interesting in the future, nagios plug-in for notifications, weather map can be interesting if is very reliable. e2emon is used for debugging. RAL: as FNAL IN2P3: performace test important, may not interested in alarms, e2emon is not in use because unreliable. CERN: perfromannce test, traceroute, alarm, decouple e2emon. Service level definition Define response time How to measure: link capacity, BER Actions: John: Complete the SLD TbD: Take Responsibility for Operations John: Lead the Operational Conference Calls Guillaume: follow up GGUS integrations of LHCOPN tickets. MonWG: define requirments for perfsonar and provide it to Geant. Next meeting 28-29 June 2010 in Barcelona Thanks to JANET for hosting the event and for the dinner.
    • 12:30 12:31
      Meeting end 1m