Report to HEPCCC March 17, 2000 Matthias Kasemann, FNAL

1
Report to HEPCCCMarch 17, 2000Matthias
Kasemann, FNAL

2
Related URLs

• ICFA-SCIC Homepage
• http//www.hep.net/ICFA/index.htmlCern -gt
  Scientific Committees -gt ICFA -gt ICFA Standing
  Committee on International Connectivity
• ICFA-NTF Homepage
• http/nicewww.cern.ch/davidw/icfa/icfa-ntf.html
• ICFA-NTF July98 Report
• http//nicewww.cern.ch/davidw/icfa/July98Report.h
  tml

3
ICFA meeting, Vancouver, 1998

• ICFA received the final report of the
  Networking Task Force (ICFA-NTF).
• Decision create a Standing Committee on
  Interregional Connectivity (ICFA-SCIC).
• Committee members represent major HEP user
  communities and laboratories.
• Focus should be on intercontinental connectivity

4
Charge to ICFA-SCIC

• Make recommendations to ICFA concerning the
  connectivity between America, Aisia and Europe.
• As part of the process of developing
  theserecommendations, the committee should
• monitor traffic,
• keep track of technology developments,
• periodically review forecasts of future bandwidth
  needs, and
• provide early warning of potential problems.
• Create subcommittees when necessary to meet the
  charge.
• The chair of the committee should report to ICFA
  once a year, at its joint meeting with laboratory
  directors.

5
ICFA-SCIC membership

• The chair is appointed directly by ICFA. ?
• Each of the major user laboratories, CERN, DESY,
  FERMILAB, KEK and SLAC, should appoint one member
  each. ?
• ECFA, DPF jointly with IPP, and ACFA, should
  appoint two members each. ?
• ICFA will appoint one member from theRussian
  federation and one member from South America. ?

6
ICFA-SCIC membership

• The representatives from the laboratories are
• Manuel Delfino, (CERN),Michael Ernst
  (DESY),Kasemann (Fermi) (chair),Yukio Karita
  (KEK),Richard Mount (SLAC).
• The North American user representatives are
• Harvey Newman (USA),Dean Karlen (Canada).
• For Russia new member

• ECFA has nominated
• Frederico Ruggieri (INFN Frascati),Denis Linglin
  (IN2P3, Lyon).
• ACFA has nominated
• Prof. Rongsheng Xu (Computer Center, IHEP
  China)
• Prof. HwanBae Park (Korea University)
• For South America Sergio F. Novaes (University
  de S.Paulo)

7
ICFA-SCIC meetings

• April 15. - 16. at FNAL.
• Main topics
• review charge to SCIC,
• review work of ICFA-NTF( a lot of it overlaps
  with SCIC charge),
• define priorities and projects,
• organize work (and subgroups).
• Video Conference on July 6, 1999.
• Topics
• update on status of network connectivity
• working group plans
• action items

8
Bandwidth Growth Observation/Prediction

• Technology Tracking and Cost Model will be
  performed by the ICFA SCIC Committee

9
Bandwidth Requirements for HENP

• Several sources of information have been used to
  estimate the bandwidth requirements
• ICFA NTF Questionnaire on Computing and
  Networking Needs
• Computing Technical Proposals and Reports
• Scaling according to computing and local area
  network (LAN) speeds
• Scaling according to the data rate to storage,
  and to the accumulated data volume
• Bandwidth available at Homes
• Bandwidth available on the Mass Market
• Bandwidth required for Physicists' Tasks
  (Reconstruction, interactive Analysis, Remote
  Collaboration,)
• Scaling according to the speed of major links
  common in national and research networks

10
ICFA Network Task Force Bandwidth Requirements
Estimate (Mbps)
See http//l3www.cern.ch/newman/icfareq98.html

11
ICFA SCIC meeting Nov. 13, 1999

• Invited UK representative Richard Hughes-Jones
• Topics discussed
• Status reports from Canada, CERN, DESY, France,
  Italy, Japan, UK.
• Wide Area networks end-to-end performance
  measurements for video and file transfer show
  need for application tuning.
• Quality of Service for National Research Networks
  and monitoring results.
• Plans for Video system (VRVS project) for higher
  quality and extended compatibility to video
  clients.

12
Impact of loss on applications

• Email
• fairly insensitive to quality, may be delayed but
  keeps retrying for days and eventually gets
  through
• Web
• usually has human but expectations are low,
  performance often more limited by server, human
  present so can retry
• Bulk file transfer
• unattended, if gt 10-12 loss connections can
  time out
• Interactive telnet, voice
• very time loss sensitive
• E.g. telnet/ssh loss of gt 3 severely impacts
  typing ability

Importance of loss/performance
13
Performance

• Measurements from
• 28 monitors in 15 countries
• Over 500 remote hosts
• 72 countries (covers all 56 PDG booklet
  countries)
• Over 1200 monitor-remote site pairs
• Over 50 of HENP collaborator sites are
  explicitly monitored as remote sites by PingER
  project
• Atlas (37), BaBar (68), Belle (23), CDF (73),
  CMS (31), D0 (60), LEP (44), Zeus (35), PPDG
  (100), RHIC(64)

14
Results Top level view - Aug-99
2000 pairs in 56 countries
packet loss between regions
Monitoring region
Remote region
Poor (2.5-5) V. poor (5-12)
Good (0-1) Acceptable (1-2.5)
Bad (gt 12)
Within region (on diagonal) good to acceptable
15
European performance from U.S.
16
E. Eur-ope Loss
Good Accept Poor V. poor
Bad
17
Russia
ESnet NSk good, ESnet ITEP IHEP improved
with new satellite Canada Edu bad all
over DESY, CERN improved to acceptable to ITEP,
IHEP, NSK with new satellite, Dubna still v.
poor to bad, UK poor to ITEP NSK KEK good to
NSk, v. poor to ITEP
18
Europe seen from U.S.
19
India / Mumbai / TIFR
Just added as beacon site
ESnet acceptable, Brazil, E. Europe/Russia poor
to bad Got better for Japan (KEK RIKEN), ESnet,
W. Europe in Oct-99, but now worse again (5 days
in Nov) Stanford vs. CMU SLAC, possibly BBN-CW
peering
20
E. Asia
Poor (2.5-5) V. poor (5-12)
Good (0-1) Acceptable (1-2.5)
Bad (gt 12)
Japan good to acceptable to N. America W.
Europe S. America poor to all E. Asia Hong Kong
China similar (v. different routes)
21
Asia seen from U.S.
22
S. America
Poor (2.5-5) V. poor (5-12)
Good (0-1) Acceptable (1-2.5)
Bad (gt 12)
Generally poor to bad Argentina is bad with
everyone Within Brazil within Colombia,
Brazil to Colombia is acceptable Columbia looks
like to be the best
23
Latin America, Africa Australasia
24
Middle East

• SLAC-Israel big change Sep 5
• RTT went from 250 msec (E3 via London) to 620
  msec
• Loss went from 6 to 1
• ESnet peers with Israel at Chicago STAR-TAP (T3
  satellite to Israel)
• 9 hops
• Iran (chapar.ipm.ac.ir)
• RTT 1000 msec.
• 11 loss
• ESnet - ATT (NY) - Unisource (NL) - archway
• 13 hops

25
Bulk transfer - Performance Trends
Bandwidth TCP lt 1460/(RTT sqrt(loss))
Note E. Europe NOT catching up
26
Problem areas - summary

• Germany was bad with .ca .edu yet good with
  ESnet. DESY improved to poor/acceptable in Aug
  with dedicated 3.5Mbps PVC to US/Canada RE,
  apart from Carleton
• Russia (W) bad to .ca .edu, good to ESnet,
  mixed to Europe, poor to .jp. Dubna worse than
  others. ITEP/IHEP better since new satellite
• Former E. block generally poor to bad
• China/Hong Kong poor to very poor with most
• S. America poor to very poor
• India poor

27
Monitoring results summary(1)

• Performance is getting better
• Within Western NRENs things are good
• Good enough even for VoIP in terms of RTT,
  jitter, loss
• Internet reliability
• in some cases is beginning to approach that
  specified in phone company frame-relay contracts,
  but still has a way to go to meet phone company
  standards of 99.999
• Improving QoS requires some combination of
• Increased bandwidth, but even keeping pace with
  growing requirements takes constant upgrades and
  investment
• Managed/reserved bandwidth works today in several
  cases
• Diff Serv has big potential but it is still a
  research topic

28
Monitoring results summary(2)

• International performance from US to sites
  outside W. Europe, Japan, Korea is generally
  poor to bad
• Transoceanic, needs special care, peering is
  critical
• E. Europe, Russia, China, India, S. America
  performance is where N. America W. Europe were
  4 years ago and may not be improving as fast so
  the discrepancy is likely to increase.

29
Report on immediate problem areas

• One particular problem area is the outlying
  regions (outside the developed regions of the
  world, W. Europe, N. America and Japan)
• who have poor connectivity (and may be falling
  further behind) and yet are delivering important
  contributions to HEP.
• In many cases the countries need an enabler to
  push the Internet connectivity for the NREN.
• ICFA members are closer to the level that can
  have an impact on the political representatives
  in the regions.

30
VRVS summary

• Principal Investigator Caltech and ESnet
• Collaborators CERN, Internet2/UCAID
• The Virtual Rooms Videoconferencing System namely
  called VRVS has been put in production since
  early 1997.
• It provides a low cost, bandwidth-efficient,
  extensible tool for videoconferencing and
  collaborative work over networks within the High
  Energy and Nuclear Physics communities and to
  some extent within Research and Education at
  large.

31
VRVS Current future

• VRVS is now a production system
• As of today, more than 1937 machines from 1253
  different users are registered into the system.
• During year 1999, 872 Multipoint Conferences has
  been Conducted (Total 2325 Hours).
• More than 3000 point to point connection
  established.
• 7 Virtual Rooms are available for World Wide
  Conferences in addition to the 4 available for
  each Continent (America only, Europe only, Asia
  only).
• VRVS Future evolution/integration (RD)
• Deployment and support of VRVS.
• High Quality video and audio (MPEG1, MPEG2,..).
• Shared applications, environment and workspace.
• Integration of H.323 I.T.U Standard into VRVS.
• Quality of Service (QoS) over the network.
• Documentation and user-configuration
  recommendations.
• Improved security, authentication and
  confidentiality.

32
ICFA-SCIC would like to point out

• How can we help to improve the network situation
  of a poorly connected and spend limited amount of
  money?
• Suggested goal
• HENP members of the country should participate in
  RD project with a western Lab/experiment that
  requires video conferencing for all collaborators
  to participate in, this will help to drive the
  requirement.
• LHC detector production quality control,
  communication to production site
• fully participate in LHC meetings and working
  groups from home institutes, publications of
  schools
• contribution to collaborative software
  development (à la VRVS) with special emphasis on
  low bandwidth requirements.

33
ICFA-SCIC would like to recommend

• Suggested goal (ctd)
• This proposal needs to be acceptable to the
  country, which requires pre-discussion with the
  country.
• Video conferencing rooms and tools must be more
  widely available (at center (CERN, DESY, FNAL,
  SLAC) and remote)
• low entrance cost for VRVS on desktop and
  conference rooms
• Establish a 2Mbps line to the country.