Data-Driven Development

The Ever-Expanding Data Universe

The rapid growth of internet users and faster network speeds is driving an avalanche of electronic data. About 3.5 billion people globally were using the internet in 2017, up 73 percent, or 1.5 billion, since 2010 (figure 2.1, panel a), and penetration has risen to almost half the world (48 percent in 2017). The rapid increase in users is driving demand for more internet content.

End-user internet speeds are also increasing rapidly, in turn driving use of broadband content and applications (figure 2.1, panel b). Global average wired broadband speeds are projected to nearly double from 25 megabits per second (Mbps) in 2015 to 48 Mbps by 2020 as more users move to fiber and higher-speed coaxial cable. As speeds rise, so does demand for video content.1 Mobile broadband speeds, which are much lower than fixed speeds, averaged just 2 Mbps in 2015. This will more than triple to 6.5 Mbps by 2020 as more users switch to fourthgeneration technologies. Mobile speeds vary greatly by device; smartphones are nearly three times faster than the global average, which results in more time spent online. In the United Kingdom, time spent on the internet more than doubled between 2005 and 2015 from 10 hours a week in 2005 to 23 hours in 2015.

Figure 2.1 Internet users and broadband speeds

Data can either be measured as stock (the amount of data stored in a location) or flow (amount of data transmitted from one location to another). One stock indicator is the number of websites providing partial information about content growth on the internet. According to Netcraft, a leading research firm covering the internet, 170 million websites were active in June 2016, up from just 8 million in June 2000. Worldwidewebsize.com puts the number of indexed web pages at 4.5 billion.2 Although useful, these numbers still lack the ability to portray the full scale of data accessible over telecommunication networks. They do not include the so-called dark web, ranging from innocuous private sites collecting sensor data to nefarious sites carrying out illegal or semi-legal activities. Furthermore, not all data going over the internet is from websites; it can also arise out of voice over internet protocol, video conferencing, gaming, and machine-to-machine communication.

More is known about data flows over the internet. In the past, separate networks existed for specific content and functions: for instance, telecommunications for voice and, later, text messages; broadcasting for television and radio; and private networks for businesses. The development of the internet and internet protocol (IP) communications has changed all that. Communications networks have generally shifted from circuit-switched to packet-switched IP networks, enabling virtually any type of content, from voice to text to multimedia, to be encoded and distributed digitally. According to information technology (IT) company Cisco, traffic over the internet will grow by more than 20 percent a year between 2015 and 2020. This data deluge has popularized a new vocabulary of petabyte and exabyte that spell checkers have not yet caught up with. Cisco proclaimed that the world entered the zettabyte era (an amount equivalent to 250 billion DVDs) in 2016 when annual global internet traffic surpassed 1 zettabyte.

It is useful to understand how internet traffic is classified to understand how devices, users, applications, and services are driving this growth. Internet traffic consists of IP and managed IP traffic. The former is exchanged between internet service providers (ISPs), whereas the latter is end to end within the same ISP's network. IP traffic can be further disaggregated by whether it emanates from fixed or mobile networks. The two accounted for three-quarters of internet traffic in 2016, with fixed making up more than 90 percent (figure 2.2, panel a). Managed IP traffic is forecast to decline by 10 percentage points between 2015 and 2020, and the share of mobile data in total traffic is projected to rise from 5 percent in 2015 to 16 percent by 2020.It is useful to understand how internet traffic is classified to understand how devices, users, applications, and services are driving this growth. Internet traffic consists of IP and managed IP traffic. The former is exchanged between internet service providers (ISPs), whereas the latter is end to end within the same ISP's network. IP traffic can be further disaggregated by whether it emanates from fixed or mobile networks. The two accounted for three-quarters of internet traffic in 2016, with fixed making up more than 90 percent (figure 2.2, panel a). Managed IP traffic is forecast to decline by 10 percentage points between 2015 and 2020, and the share of mobile data in total traffic is projected to rise from 5 percent in 2015 to 16 percent by 2020.

Figure 2.2 Global IP traffic and global consumer IP traffic

Businesses and consumers generate traffic, with the latter accounting for more than 80 percent in 2015, a share not projected to change much through 2020. Video dominates consumer IP traffic. It accounted for 38 exabytes a month of traffic in 2016, 71 percent of consumer IP traffic, and 43 percent of total IP traffic. It is forecast to grow by more than 30 percent a year so that, by the year 2020, it will account for 82 percent of consumer traffic and 57 percent of total IP traffic. Online gaming is projected to be the fastest-growing traffic stream between 2015 and 2020, at 47 percent per year. However, it accounts for a tiny share of total consumer traffic and its contribution will only rise from 0.2 percent to 0.4 percent.