african internet performance: how bad is it; what can be done?
DESCRIPTION
eGY. African Internet Performance: How bad is it; what can be done?. Prepared by: Les Cottrell SLAC , Umar Kalim SEECS,NUST/SLAC IHY-Africa/SCINDA 2009, Livingstone, Zambia, 7-12 June 2009. www.slac.stanford.edu/grp/scs/net/talk09/ihyjun09.ppt. Summary. African Infrastructure - PowerPoint PPT PresentationTRANSCRIPT
1
Prepared by: Les CottrellSLAC,Umar KalimSEECS,NUST/SLAC
IHY-Africa/SCINDA 2009, Livingstone, Zambia, 7-12 June 2009
www.slac.stanford.edu/grp/scs/net/talk09/ihyjun09.ppt
AfricanInternet Performance:
How bad is it; what can be done?
eGY
2
Summary• African Infrastructure
• Methodology of measuring Internet performance
• Overall world Internet performance & where does Africa stand
• Africa directions– Wireless/fibre, Routing, Costs, Difficulties,
• Conclusions & further information
Africa is Huge• Hard to get coverage (e.g. fibre) everywhere
3
• India 10% area, but > population
…and diverse (e.g. languages)• Resources, religions,
geography …• More than 1,000 indigenous
African languages including several spoken by tens of millions such as Igbo, Swahili, Hausa, Amharic, and Yoruba;
• Plus Arabic, English, French, Portuguese, Afrikaans, Spanish, Indian languages, others
4
5
African World Status
• Internet city connections
Fibres
Light at night
CapacityFrom Telegeography
Sub-Saharan broadband costs off-scale
6
www.itu.int/ITU-D/ict/publications/idi/2009/index.html
Source ITU
1 yr of Internet access > average annual income of most Africans,
Survey by Paul Budde Communications
7
Why Make Internet Measurements?• In the Information Age Information Technology (IT)
is the major productivity and development driver., particularly science & education
• Travel & the Internet have made a global viewpoint critical
• One Laptop Per Child (“$100” computer) – New thin client paradigm, servers do work, requires
networking (Google: “Negroponte $100 computer”), driving Intel & AMD cheap net-books,
– Internet enabled Smart phones (e.g. iPhone)– Enables “Internet Kiosk & Cafe” can make big difference
• So we need to understand and set expectations on the accessibility, performance, costs etc. of the Internet
8
PingER Methodology extremely Simple
Internet
10 ping request packets each 30 mins
RemoteHost(typicallya server)
Monitoring host
>ping remhost
Ping response packets
Measure Round Trip Time & Loss
Data Repository @ SLAC
On
ce a Day
Uses ubiquitous pingICTP
9
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– ICFA/SCIC
• Most extensive E2E Active Internet Measurement
– Monitors (>40 in 23 countries – 3 Africa)
– Beacons ~ 90– Remote sites (~740)
• >165 countries (98% world’s population, >99% world’s connected population)– >45 countries in Africa
10
World Measurements: Min RTT from US• Min RTT indicates best possible, i.e. no queuing• >400ms probably geo-stationary satellite (red & magenta)• Maps show increased coverage by fibre (less GEOS)• Only a few places still using satellite for international
access, mainly Africa & Central Asia
2000 2008
Loss
11
With TCP (>80% Internet traffic) recovery from loss can take several seconds, such delays make interactive use annoying to impossible.For non TCP multi-media traffic loss causes poor voice/video (VoIP/H323) above 1.5%,loss > 0.5% unacceptable for IPTVhttp://www.slac.stanford.edu/comp/net/wan-mon/tutorial.html#loss
Africa by far worst region,
10-20 times worse than developed regions
12
World Throughput Trends Behind Europe5 Yrs: Russia, Latin America, Mid East 6 Yrs: SE Asia9 Yrs: South Asia12 Yrs: Cent. Asia16 Yrs: Africa
In 10 years at the current rate Africa will be 1000 times
worse than Europe
Derived throughput ~ 8 * 1460 /(RTT * sqrt(loss))Mathis et. al
1993
13
Some Other World Views
Voice & video (de-jitter) Network & Host Fragility
Data TransferCapacity
14
Mediterranean. & Africa vs HDI
• There is a good correlation between the 2 measures• N. Africa has 10 times poorer performance than Europe• N. Africa several times better than say E. Africa• E. Africa poor,
limited by satellite access
• W. Africa big differences, some (Senegal) can afford SAT3 fibre others use satellite
• Great diversity between & within regions
HDI related to GDP, life expectancy, tertiary education etc.
15
Opportunities: Routing• Seen from TENET
Cape Town ZA
• Only Botswana & Zimbabwe are direct
• Most go via Europe or USA
• Wastes costly international bandwidth, subsidizes international carriers
• Need IXPs in Africa
Opportunities: Fibre, satellite, mobiles• Satellite is extremely effective in reaching places where the volume of traffic
would not justify a fibre connection. • But GEOS satellite $/Mbps 300-1000 x Fibre, severely bandwidth-constrained
and high latency• So fibre international and to major cities
– Scramble to provide international fibre for
World Cup 2010– then wireless (cell phone, wimax, …) – cell phone growth leads Internet growth by 4.5 years
• 16 LEOS (reduce latency) - Sep 2008• Google signed up with Liberty Global and HSBC in a bid to launch 16 LEOS satellites, to bring high-
speed internet access to Africa by end 2010
– ABUJA Africa's first communications satellite suffered an energy failure just 18 months after its launch - reported Nov. 2008
16
17
African International Fibres 2010
Current:SAT-3-WASC run by a consortium of state monopolies that has opted for elite rather than mass market.Prices tend to align to satellite in the absence of competition!
“Black” Fibres installedalong roads, pylons etc. remain unused because of monopoly regulation!
Near Future: driven by World Cup in 2010
18http://www.internetworldstats.com/
What else is driving it
Huge growth
~ 3x lower penetration than any other regionhuge potential market
Many systemic factors:Electricity, import duties,skills, disease, protectionist policies, conflict, corruption.
19
Conclusions: The bad• Poor performance affects data transfer, multi-media,
VoIP, IT development & country performance / development
• DD exists between regions & countries, rural vs cities, poor vs rich, old vs young…
• Decreasing use of satellites, expensive, but still needed for many remote countries in Africa and C. Asia
• Last mile problems, and network fragility• Current providers (cable and satellite) have a lot to loose
– Many of these have close links to regulators and governments (e.g. over 50% of ISPs in Africa are government controlled)
• Africa worst by all measures (throughput, loss, jitter, DOI, international bandwidth, users, costs …) and falling further behind.
Conclusions: There is Hope• World cup: international fibre access + competition• LEOS• Leapfrog last mile fixed wire with wireless• Cheaper end points: OLTP, netbook, smart-phones
• Banding together of universities => leverage influence & get deals => NRENs => IXPsUsers– E.g. Ubuntunet, Bandwidth Initiative
• Standards:– Harmonization of regulations country to country– Cheaper cell phone, can’t afford multiple technologies & frequencies
• Regulatory regimes becoming:– more open/transparent, less resistant to change
20
“The way we develop here in Africa will be different from the way the big nations
developed. They grew up with computers. We are growing up with mobile phones.
- Fritz Ekwoge”
Conclusions: PingER• Quantitatively Measures Internet performance
– non subjective, – relatively easy/quick to measure (c.f. ITU etc methods)
• So monthly, daily updates
– correlates strongly with economic/technical/development indices– Increase coverage of monitoring to understand Internet performance
– Lots of granularity:• within countries, monthly, daily
• Gives baselines, trends, effect of improvements
• Relative comparisons countries, regions, sites
• Good coverage for Africa– Need: Chad, Comoros, Eq. Guinea, Sao Tome, Somalia
21
22
More Information
• Thanks: – Incentive: ICFA/SCIC, Monique Petitdidier, ICTP, ITU – Funding: DoE/SLAC/HEP, Pakistan HEC– Effort: SLAC, NUST, ICTP (Trieste), FNAL, Georgia Tech,
administrators at over 40 monitoring sites in 23 countries• ITU/WIS Report 2006 & 2007 (or Google: “WSIS Report 2007”)
– www.itu.int/osg/spu/publications/worldinformationsociety/2007/report.html– www.itu.int/ITU-D/ict/publications/idi/2009/index.html
• Higher Education in Sub-Saharan Africa– www.arp.harvard.edu/AfricaHigherEducation/Online.html
• PingER– www-iepm.slac.stanford.edu/pinger, sdu.ictp.it/pinger/africa.html– www.slac.stanford.edu/xorg/icfa/icfa-net-paper-jan09/
• Global Information watch: www.giswatch.org • Need network contacts in Africa:
23
Extra Slides
24
Trends:Losses
• N. America, Europe, E. Asia, Oceania < 0.1%
• Underdeveloped 0.3- 2% loss, Africa worst.
• Mainly distance independent
• Big impact on performance, time outs etc.
• Losses > 2.5 % have big impact on interactivity, VoIP etc.
25
• ~ 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
Trendlines for IPDV from SLAC to World Regions
N. America E. Asia
Europe
Australasia
S. Asia Africa
Russia
L. America SE Asia
C Asia
M East
Usual division into Developed vs Developing
Jitter
26
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)– Africa & C. Asia not possible, S. Asia with patience OK
MOS of Various Regions from SLACImprovements very clear, often due to move from satellite to land line.Similar results from CERN (less coverage)
Usab
le
27
Leading African Countries
Country Population
[Sort] Int'l BW Mbps
Int'l BW / capita (bps)
Internet Users
Internet users/ 1000 capita
BW (bps)/ Internet User
DigitalOpportunityRank
Egypt 82,073,660.00 3784 46.105 1000000 12.1842 3784 90
South Africa 43,743,316.00 881.5 20.152 1012500 23.1464 870.617 91
Senegal 12,938,350.00 775 59.899 19351 1.49563 40049.6 112
Cameroon 18,569,348.00 155 8.3471 6500 0.35004 23846.2 137
Nigeria 139,070,856.00 150 1.0786 350000 2.5167 428.571 155
Kenya 38,213,024.00 113.39 2.9673 80000 2.09353 1417.38 164
Uganda 31,621,980.00 100 3.1624 8000 0.25299 12500 152
Burkina Faso 14,866,133.00 76 5.1123 14238 0.95775 5337.83 163
Cote d'Ivoire 18,465,326.00 55.42 3.0013 13747 0.74448 4031.43 144
Benin 8,349,959.00 47 5.6288 6396 0.76599 7348.34 147
Niger 13,364,797.00 30 2.2447 3117 0.23322 9624.64 179
Mozambique 21,379,584.00 18.5 0.8653 25000 1.16934 740 169
Ethiopia 78,697,922.00 10 0.1271 12155 0.15445 822.707 173
Namibia 2,067,433.00 9 4.3532 19000 9.19014 473.684 109
28
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
US & CanadaEurope
E Asia
C Asia
SE Europe
SE Asia
S AsiaOceania
Africa
L America M East
Russia
DevelopedRegions
DevelopingRegions
29
Throughput• Derive from:
Thru ~ 8 * 1460 _____________(RTT * sqrt(loss))
30
African Situation
• Access to the internet is so desirable to students in Africa that they spend considerable time and money to get it. Many students surveyed, with no internet connection at their universities, resorted to private, fee-charging internet cafes to study and learn. www.arp.harvard.edu/AfricaHigherEducation/Online.html
Internet Café in Ghana
• School in a secondary town in an East Coast country with networked computer lab spends 2/3rds of its annual budget to pay for the dial-up connection.
– Disconnects Heloise Emdon, Acacia Southern Africa
1 yr of Internet access > average annual income of most Africans, Survey by Paul Budde Communications
• Survey (IHY meeting Ethiopia in November ’07) of leading Universities in 17 countries (will repeat with more clarity):
– Each had tens of 1000’s of students, 1000 or so staff– Best had 2 Mbits, worst dial up 56kbps– Often access restricted to faculty
PingER: African coverage• Host monitored in 50 of ~60 countries (98.7% pop)• 131 hosts monitored in Africa• Cannot find hosts in Chad, Comoros, Eq. Guinea, Sao Tome, Somalia
• Yellow only 1 host (so could be anomalous, e.g. Libya)
• Need help for contacts: ([email protected])
31
PingER Sites vs IHY sites
32
Magnetometer
PingER sites
SID/GPS
• IHY sites with good Internet access nearby may be able to use it to transfer data or even control
• IHY Coordinates from Monique Petitdidier (CNRS), Deborah Scherrer (Stanford), Barbara Thompson (NASA)