Quantifying the need for Improved Network Performance for S. Asia

1. Quantifying the need for Improved Network Performancefor S. Asia Prepared by: Les CottrellSLAC& Shahryar KhanNIIT For the Internet2 Special Interest Group on Emerging NRENs Planning Meeting: Enhancing Research and Education Connectivity to and within South Asia April 26, 2007 http://www.slac.stanford.edu/grp/scs/net/talk07/sasia-case-apr07.ppt

2. How do we Measure Network Performance?

3. PingER Methodology Uses ubiquitous ping >ping remhost Remote Host (typically a server) Monitoring host Internet 10 ping request packets each 30 mins Once a Day Ping response packets Data Repository @ SLAC Measure Round Trip Time & Loss

4. PingER Deployment • PingER project originally (1995) for measuring network performance for US, Europe and Japanese HEP community - now mainly R&E sites • Extended this century to measure Digital Divide: • Collaboration with ICTP Science Dissemination Unit http://sdu.ictp.it • ICFA/SCIC: http://icfa-scic.web.cern.ch/ICFA-SCIC/ • >120 countries (99% world’s connected population) • >35 monitor sites in 14 countries • Monitor 44 sites in S. Asia • Most extensive active E2E monitoring in world

5. Where does S. Asia fit Compared to Rest of World

6. World Measurements: Min RTT from US • Maps show increased coverage • Min RTT indicates best possible, i.e. no queuing • >600ms probably geo-stationary satellite • Between developed regions min-RTT dominated by distance • Little improvement possible • Only a few places still using satellite for international access, mainly Africa & Central Asia 2000 2006

7. Unreachability • All pings of a set fail ≡unreachable • Shows fragility, ~ distance independent • Developed regions US, Canada, Europe, Oceania, E Asia lead • Factor of 10 improvement in 8 years • Africa, S. Asia followed by M East & L. America worst off • Africa NOT improving SE Asia L America M East C Asia Oceania S Asia SE Europe Russia Developing Regions Africa E Asia Developed Regions US & Canada Europe

8. Losses • Mainly distance independent • Big impact on performance, time outs etc. • Losses > 2.5 % have big impact on interactivity, VoIP etc. • N. America, Europe, E. Asia, Oceania < 0.1% • Underdeveloped 0.3- 2% loss, Africa worst.

9. Jitter • ~ Distance independent • Calculated as Inter Packet Delay Variation (IPDV) • IPDV = Dri = Ri – Ri-1 • Measures congestion • Little impact on web, email • Decides length of VoIP codec buffers, impacts streaming • Impacts (with RTT and loss) the quality of VoIP Usual division into Developed vs Developing Trendlines for IPDV from SLAC to World Regions C Asia Russia S. Asia Africa SE Asia L. America M East Australasia Europe N. America E. Asia

10. VoIP & MOS • Telecom uses Mean Opinion Score (MOS) for quality • 1=bad, 2=poor, 3=fair, 4=good, 5=excellent • With VoIP codecs best can get is 4.2 to 4.4 • Typical usable range 3.5 to 4.2 • Calc. MOS from PingER: RTT, Loss, Jitter (www.nessoft.com/kb/50) MOS of Various Regions from SLAC Improvements very clear, often due to move from satellite to land line. Similar results from CERN (less coverage) Usable

11. World thruput seen from US Throughput ~ 1460Bytes / (RTT*sqrt(loss)) (Mathis et al) Behind Europe 6 Yrs: Russia, Latin America 7 Yrs: Mid-East, SE Asia 10 Yrs: South Asia 11 Yrs: Cent. Asia 12 Yrs: Africa South Asia, Central Asia, and Africa are in Danger of Falling Even Farther Behind

12. Normalized for Details • Note step changes • Africa v. poor • S. Asia improving • N. America, Europe, E Asia, Oceania lead

13. “Development” Indices • There are many “development” indices today (values 0-1), e.g.: • UNDP Human Development Index (2006, 177 countries) • UNDP Technology Achievement Index (2001, 72 countries) • ITU Digital Access Index (2003) and the Digital Opportunity Index (2006), both 180 countries • World Economic Forum’s Network Readiness Index (2004, 2005, 2006-2007: 122 countries) • Harvard University Network Readiness Index (2002, 75 countries) … • Typically subset of: GDP/capita, knowledge (e.g. tertiary education enrollment), life expectancy, network (hosts/capita, access, policy, usage, affordability, users/capita); technology (patents, royalties, exports, phones/capita, electricity) • The size of the Internet infrastructure is a good indication of a country's progress towards an information-based economy. • Indices are hard to gather, agree on, many countries do not report • Most Internet traffic in a developing country is international (75-90%) • We measure international Internet performance which is an interesting (good?) indicator.

14. Digital Access Index (DAI) infrastructure, affordability, knowledge and quality and actual usage of ICTs • Most European countries > 1500 Kb/s throughput and greater than 0.6 DAI. Exceptions: • Malta, Belarus and Ukraine. • Balkans is catching up with Europe, exception Albania is way down. • E. Asia apart from China clusters • M East: Israel & Cyrus close to Europe, Iran way down • SE Asia 3 cluster: Singapore at top, Malaysia and Brunei middle, Vietnam & Indonesia at bottom • S. Asia 2 clusters: • India, Pakistan, Sri Lanka • Bangladesh, Bhutan, Nepal • Africa at bottom • Correlation strong

15. Focus on S. Asia

16. S. Asia Coverage Min-RTT from CERN • Monitor 44 hosts in region. • 6 Monitoring hosts Loss from CERN

17. Routing • Between developing countries often use transcontinental links (like Europe in 80’s), e.g.: • Pak to Pak or India to India is direct, however, • Between Pak & India via US or Canada or Europe • Between India or Pak and Bangladesh or Sri Lanka via US or UK • India=>UK=>India (Delhi=>Mumbai)=>Nepal • India=HK=India=>Bhutan • Wastes costly transcontinental bandwidth • Drastically extends RTT & degrades performance • Need International eXchange Points (IXPs)

18. Bandwidth & Internet use • Note Log scale for BW • India region leader • Pakistan leads bw/pop • Nepal very poor Bit/s • Pakistan leads % users • Sri Lanka leads hosts%% • Pakistan leads bw/pop • Nepal, Bangladesh, Afghanistan very poor

19. S Asia MOS & thruput Daily throughputs from US to S Asia • weekend vs. w’day, day vs night = heavy congestion RTT NIIT to QAU Pak (1 week) RTT ms Fr Su Mo Tu Sa We Th Mean Opinion Score to S Asia from US Usable • Last mile problems • Divides into 2 • India, Maldives, Pakistan, Sri Lanka • Bangladesh, Nepal, Bhutan, Afghanistan Pakistan

20. VC from this meeting • They can be made to work, they are exceedingly valuable, takes expertise/support, may not use Internet, Internet is cheap.

21. DAI vs. Thru & S. Asia • More details, also show populations • Compare S. Asia with developed countries, C. Asia

22. Conclusions • DD exists between regions, within regions, within countries • S Asia divides into two • Applications fail, no connectivity, telnet, VoIP/multimedia, Grid clusters and data transport (e.g. Pakistan) • Decreasing use of satellites, expensive, but still needed for many remote countries in Africa and C. Asia • Last mile problems, and network fragility • Affects data transfer, multi-media, VoIP • Internet performance (non subjective, relatively easy/quick to measure) correlate strongly with economic/technical indices • Increase coverage of monitoring to understand Internet performance • Need funding (used to be DoE (research in net mon), SLAC, US State Dept, HEC/MOST Pak), Pak continues, US needs to match

23. More information/Questions • Acknowledgements: • Harvey Newman and ICFA/SCIC for a inspiration, I2 for this meeting and immediate demand; NIIT/Pakistan, Maxim Grigoriev (FNAL), Connie Logg (SLAC), Warren Matthews (GATech) for ongoing code development for PingER; USAID MoST/Pakistan for development funding, SLAC for support for ongoing management/operations support of PingER • PingER • www-iepm.slac.stanford.edu/pinger, sdu.ictp.it/pinger/africa.html • Case Studies: • https://confluence.slac.stanford.edu/display/IEPM/South+Asia+Case+Study • https://confluence.slac.stanford.edu/display/IEPM/Sub-Sahara+Case+Study • http://sdu.ictp.it/lowbandwidth/program/case-studies/index.html