A decadelong quest aims to pinpoint the Internet bottlenecks holding Africa back
The University of Kinshasa, the largest university in the Democratic Republic of Congo, has nearly 30 000 students, faculty, and research staff—but only 800 computers. And although its internal data network is fast enough to support on-campus e-mail, virtual library access, and online coursework, its link to the outside world is no better than that of a typical household in the United States or Europe.
When I visited the university about a year ago, the restricted capacity meant that only 200 senior staff had Internet privileges—for whatever those were worth. I was able to go online for a short time, but the connection was barely usable. Web pages timed out or loaded so slowly that after about 10 minutes, I simply gave up.
Regrettably, the situation in Kinshasa is familiar to millions more schools, organizations, and communities in emerging parts of the globe. And the technical gap between developed and developing regions only widens the economic one. Without these connections, many developing countries miss out on innovations that offer higher standards of living, such as telemedicine, remote learning, and online commerce.
Nowhere is this disparity felt greater than in Africa. Less than 16 percent of Africans have access to the Internet, according to the latest market research. Compare that with 63 percent of Europeans and 79 percent of North Americans. Even in Asia, where Internet usage ranks second to last among the seven major world regions, the penetration rate is almost twice the rate in Africa.
Part of the problem is poor network performance. The total bandwidth available to shuttle data between African countries and the rest of the world in 2011 was less than 1 terabit per second. That’s about one-seventieth of Europe’s international bandwidth capacity. Making matters worse, the price for bandwidth in Africa is cripplingly high. While a university in Germany might pay about US $4000 per month for 1 gigabit per second of bandwidth, a school in Kenya can expect to pay $200 000 for the same service.
Yet while Africans continue to face huge technological disadvantages, conditions have nonetheless improved dramatically over the past few years. Since 2009, Africa has seen unprecedented upgrades to its broadband infrastructure. Remarkably, one of the biggest drivers was a single popular event: the 2010 FIFA World Cup. The celebrated tournament attracted more than 3 million football fans to Johannesburg that year. It also brought television crews and newscasters, whose telecommunications needs greatly exceeded what South Africa could have provided just two years before.
In 2008, Africa had only three fiber-optic links to the global Internet: two in the north and one in the west. But by the first match of the World Cup in June 2011, two additional international subsea cables had made landings up and down Africa’s eastern and western shorelines, and three more would be completed before the end of the year. Since then, telecoms, governments, and aid organizations have invested billions in additional submarine cables and terrestrial networks across the continent.
So has Internet availability improved in landlocked as well as coastal countries? Or is it merely falling behind at a slower rate than before? Where besides the University of Kinshasa are there still significant bandwidth bottlenecks? To answer these questions, we need a cheap, reliable way to measure Internet performance at various places throughout Africa. And we need to compare the results with data from other regions and track changes over time.
That’s exactly what the Ping End-to-end Reporting, or PingER, project does. I lead the project at the SLAC National Accelerator Laboratory, in California, with help from a handful of students and colleagues at the National University of Sciences and Technology (NUST) in Pakistan and at the International Centre for Theoretical Physics (ICTP) in Italy. But PingER’s reach is truly global.
Using the simple and common “ping” test, we regularly measure how well data is flowing, if at all, between pairs of hosts—typically Web servers at distant universities. Since PingER’s start nearly two decades ago, we have set up close to 100 monitoring hosts around the globe, most of which now collectively observe about 900 target hosts in 164 countries. The data are delivered daily to three archive sites, in the United States and Pakistan, and made available online.
These measurements paint a fascinating history of the Internet’s growth as seen from many different vantage points. They also reveal the impact on Internet performance of major world events, such as the 2011 uprisings in Egypt and Libya, and of major infrastructure upgrades, such as when many sub-Saharan countries switched from satellite to terrestrial links in 2009 and 2010.
Ultimately, by quantifying the digital divide between world regions and across time, we can better understand where and how to bridge it.
Ping’s ubiquity and ease of use make it especially well suited to widespread Internet monitoring, particularly in underdeveloped regions such as Africa where more advanced applications may be impractical.
The inventor of the tool, the late American computer scientist Mike Muuss, remembered writing the original code as “a little thousand-line hack” during a single evening in 1983 to troubleshoot “odd behavior” on the computer network at the U.S. Army’s Ballistic Research Laboratory, in Maryland. His program sent a small data packet known as an echo request to an Internet Protocol address, typically a remote server or network node. If the target address was reachable, it sent back the same data, and the program recorded the time it took for the round-trip journey. The echoing action of the data probe reminded Muuss of the percussive sound pulse that sonar systems use to detect objects underwater, and so he named the program after that sound—ping.
The ping program can tell us a lot about the health of an Internet connection. By sending several echo requests in short succession, we can determine each packet’s round-trip time (latency), the variability of these times (jitter), and the percentage of packets that never return (loss). High jitter or high loss typically indicates that a network path is heavily congested or there isn’t enough bandwidth to handle the traffic.
When I first started PingER in 1995, I had no intention of using ping to address network deficiencies in Africa. As the head of networking at SLAC, I set up the system simply to test connections between the laboratory and several dozen research institutions worldwide that were collaborating on BaBar, a physics experiment whose aim was to study properties of subatomic particles. To analyze results, the BaBar collaborators had to share enormous data sets, which they had to be able to transfer quickly and reliably. PingER let me keep tabs on how parts of the network were performing and root out any problems.
Over the next half decade, as word of PingER’s value spread, I extended monitoring to hundreds more physics laboratories and science centers across the globe. But the project didn’t take a humanitarian turn until 2001.
That year, while visiting the ICTP in Italy, I met two staff physicists—Hilda Cerdeira, who grew up in Argentina, and Enrique Canessa, who hails from Chile. Bolstered by the ICTP’s mission to bring first-class science and technology to developing countries, they were helping many communities, mostly in Africa, to build computer labs and wireless networks. They told me they wanted to know how well the networks were working, and they thought PingER was the perfect tool for the job. They offered to help expand the project to those parts of the world that needed it most. Within the next year, we began establishing monitoring and target hosts in countries as diverse as Bhutan, Ecuador, Jordan, and Rwanda.
I soon got my first real glimpse of just how much of a difference PingER could make. In 2004, I started working with engineering students at Pakistan’s NUST to help them set up a PingER monitoring site to assess performance on the then year-old Pakistan Educational Research Network (PERN). The network’s providers and backers touted its bandwidth of 155 megabits per second—impressive at the time. But PingER revealed that the “last mile” links to universities were dreadful. These bottleneck connections funneled data at no more than 1 Mb/s, causing long delays and high packet loss.
During one of my many visits to the university, I presented our findings to the chairman of Pakistan’s higher education commission, Atta-ur-Rahman, who was preparing to fund the next major upgrade to PERN. He clearly took PingER’s lessons to heart. When construction of PERN2 began in 2009, its plans included extending high-speed, 1-Gb/s data links all the way to university data centers.
Looking at PingER data from Africa, it’s clear that although Internet performance is improving, it still lags the rest of the world. Packet loss, for example, remains higher than in any other region except Central Asia—which includes Kazakhstan, Tajikistan, and three other republics of the former Soviet Union. What’s encouraging, however, is that loss rates in Africa have begun to drop below 1 percent in the past couple of years. The threshold is important because it means that many connections can or may soon support real-time communication services, such as video streaming and Voice over Internet Protocol (VoIP) calls.
Round-trip latency has similarly improved in recent years, most dramatically in African countries that have switched from satellite to terrestrial links. Many of the upgrades followed the landing of new submarine cables prior to the World Cup in 2010. Not only did the cables vastly expand international data capacity on the continent, they also introduced competition, which drove down the price of data transmission. Through PingER, we can see that since 2008, more than a dozen sub-Saharan countries have mostly abandoned satellite Internet, after which the latency of their connections suddenly dropped by roughly half [see map, "Football and Broadband"]. These upgrades have helped establish several high-speed national and regional research and education networks in Africa, similar to Internet2 in the United States or the Géant network in Europe.
PingER data can also be used to assess the quality of a voice connection, which telephone companies originally determined by placing callers in a quiet room and having them rate audio quality from 1 (worst) to 5 (best). For a VoIP network, the International Telecommunication Union has established a standard for calculating this mean opinion score, or MOS, based on latency, loss, and jitter. A network with an MOS greater than 3.5 can generally support VoIP services such as Skype.
By tracking MOS results over time, we can identify the approximate date when Internet-based calling emerged between one region and another. For example, it appears that VoIP calls between North America (at SLAC) and Latin America were possible beginning in 2003. Similarly, VoIP calls to the Middle East were first possible in 2004, and calls to South Asia started in 2009. Not coincidentally, 2009 was also the first year that I and my student collaborators in Pakistan began using Skype for conference calls. I remember the event fondly because it meant I could stop paying $1 per minute for a regular long-distance connection.
Africa’s Internet still has a way to go before most communities there can take advantage of these savings. In fact, Africa remains the only region whose average MOS score hasn’t yet surpassed 3.5. I suspect, however, that considering recent progress in other measures, this will begin to change in the next few years.
Another useful performance metric is throughput, a function of loss and round-trip latency that approximates the rate at which data is delivered between pairs of hosts. Given Africa’s recent improvements in both loss and latency, it’s no surprise that its throughput is increasing. But until recently, it wasn’t increasing fast enough to catch up with the rest of the world. In fact, Africa’s throughput in 2009 was comparable to Europe’s 15 years ago, and PingER data suggested the gap would widen to 25 years by 2020.
Happily, throughput has grown dramatically since 2009, putting Africa just 14 years behind Europe. If growth continues at the current rate, Africa may even catch up by 2030.
That’s a big if, though. It remains to be seen whether this uptick in growth is just a short-lived aftereffect of the 2010 World Cup or whether new investments will sustain progress in the long run.
If the future of information technology in Africa seems uncertain, that’s because it is. The 2010 World Cup undoubtedly set in motion big changes in the region. And as the continent continues to attract big investors and cutting-edge science research projects, its Internet performance will advance further still. For example, the Square Kilometre Array, a massive radio telescope being jointly built in South Africa and Australia, will haul more data between the two continents than travels through today’s entire global Internet.
The reality, however, is that most Africans still face substantial, and often unpredictable, obstacles to Internet access, including poverty, political corruption, and the lack of basic services such as power and water. Even in the more stable countries, such as Angola and Kenya, that have built hefty broadband backbones, engineers have yet to find a tried-and-true way to extend cheap, dependable Internet service to rural communities.
As PingER data show, there is plenty of work to be done. My hope is that the technology will bring awareness to overlooked shortcomings in Africa’s communications networks, as it did in Pakistan. And by expanding traditional cable networks as well as embracing “leapfrog” technologies such as Wi-Fi, low Earth orbit satellites, and mobile phones, providers will be able to deliver affordable broadband access to even the most remote corners of the continent.
This article originally appeared in print as "Pinging Africa."
For more information about how PingER data can be used to evaluate a country’s progress toward becoming an information-based society, see “A Simple Tool for Measuring Digital Development.”
About the Author
In his 43-year career, R. Les Cottrell has racked up a number of firsts. He worked with the Nobel Prize–winning team that discovered the quark. He helped establish the first Internet connection to China and worked on the first website in North America. Now at the SLAC National Accelerator Lab, in California, he runs the first global Internet monitoring system.