Over the past weeks we closely followed the impact of economic lockdowns on the ICT infrastructure to identify weaknesses and areas for future investments. Our analysis is multi-dimensional given the complexity of the ICT infrastructure which combines different types of access and backbone networks. This heterogenous infrastructure varies widely by geography. Our assessment led us to identify areas for future investments and evolve regulatory policies in both developed and developing economies. However, we also uncovered areas that we need to be particularly careful about. One of those areas is balancing competition against investments in infrastructure and the role of regulators. We review a few key observations on traffic trends which drives infrastructure architecture and investments.
Traffic Diffusion
Traffic patterns drive ICT infrastructure architecture and investments. The global economic lockdowns changed normal traffic patterns leading the infrastructure to experience different stress points. Traffic diffusion is what best describes the consequences of these lockdowns:
- Diffusion over geography: Traffic shifted away from the business and urban cores as people stayed home. This raised demand on suburban and rural networks while it reduced the load on networks in the urban core (especially the mobile networks).
- Diffusion over time: The demand for Internet is spread over longer time as work-life balance is altered. For example, morning traffic on fixed networks saw significant increase.
The main consequence of traffic diffusion is amplification of weaknesses [side note: Covid-19 is an amplifier of weaknesses in every domain!]. Changing traffic patterns exposes the weakest links of the infrastructure. This has multiple facets such as the divide between rural and urban areas and between developed countries and developing ones, and between the connected and under-connected. This latter element was visible around the world even in developed countries where in one example school children has to access the Internet from the outside of their closed school buildings.
Traffic Rebalancing
Economic lockdowns led to rebalancing of traffic across networks. Primarily, traffic shifted away from mobile networks onto fixed access networks, including indoors Wi-Fi. As a result, traffic patterns on mobile networks came to resemble those for weekends in distribution with different peak to average ratios and busy hours. Voice traffic on mobile networks picked up and mobile data served to backup to fixed networks when the latter became congested.
However, we cannot generalize these observations across the entire world where fixed access networks are weak in developing countries in terms of penetration and speed. With the mobile networks already under stress, people in many developing countries had not ‘digital’ alternatives to some basic services such as health, banking and other services. It is in these countries that the limited access to the Internet had the most adverse impact.
Traffic Type and Persistence
Traffic increased as extraneous condition changed the demand over time and geography. Data from different countries shows that most of the increase in traffic came after schools were closed while a more moderate increase happened when businesses closed [1]. Business traffic, including video conferencing, is not the primary driver of traffic. In fact, services such as Netflix and YouTube reduced their streaming bit rate which helped in saving bandwidth [20%-45% reduction in streaming rate depending on resolution of video]. Other services, such as Zoom and Webex, dynamically adapted the rate according to potential congestion feedback.
Networks held up reasonably well to traffic increase because they are designed to support busy hour traffic. Stay-at-home measures extended the busy hour traffic to longer periods. But investments in infrastructure requires traffic to be persistent – i.e. sustained demand over time. This raises the question of how persistent the new traffic patterns are? Would they revert to their norm as economies open? How quickly would that take and what is the residual impact?
Areas for Investments
Networks have largely succeeded in accommodating the change in traffic patterns. The Internet behaved the way it supposed to behave. Some of the areas where our analysis framework points don’t come as a surprise:
- ICT backbone and core Internet infrastructure. The core network experiences the aggregated demand from the access network. Investments in the backbone infrastructure addresses such influx could be well justified.
- Fixed access networks. Developing countries which focused investments on mobile infrastructure found themselves exposed with weak fixed access networks as mobile capacity could not cope with the additional traffic. We also see shortcomings in developed markets, especially in remote areas.
- Addressing the ‘under-connected’. This is an omni-present challenge specifically highlighted by the pandemic. For example, providing connectivity to children to continue their schooling or to employees to work from home became a challenge in many parts of the world – even in developed markets that have good broadband penetration.
- Network infrastructure distribution. Addressing how the telecom and cloud infrastructure handles massive traffic redistribution is a key area. Much of the cloud infrastructure is hosted in developed countries. Developing nations found that they were stranded with limited ability to compete for access to resources. Addressing cloud infrastructure on a regional and national basis is a critical issue that we expect to gain in prominence.
The Threat to Competition
While investment in infrastructure was a primary motivation for our research, we came to realize a potential threat that the pandemic and subsequent lockdowns could have on competitiveness. We see the strong possibility – which we believe has already started – of incumbent telcos and cloud companies use their dominant market position to push against new market entrants. They argue the need for a great amount of investment to build strong ICT infrastructure. Consequently, they would lobby the regulators against opening the market to new service providers which would threaten their ability to invest in infrastructure.
Balancing competition against incumbency in ICT infrastructure regulations will take on a new meaning going forward. We expect that the battle lines will harden as incumbents fight for more power. Incumbents, who have the advantage of asymmetrical knowledge of their network capabilities and costs, are well prepared to fight such a battle. This calls on regulators and the public to be ever more vigilant.
Concluding Thoughts
The pandemic highlighted the critical nature of the ICT infrastructure: investors now realize that access to the Internet is just as important as access to water, electricity, and roads. However, the investment approach to ICT infrastructure is not the same as that for other critical infrastructure. Telcos design and build networks to meet traffic demand leveraging economies of scale. Where economics are weak, so is the telco infrastructure. The pandemic served to expose this fact and its consequences. Addressing this weakness will require innovative solutions, including new business and investment models to build ICT infrastructure. This applies to developing countries which now realize that ICT infrastructure is a pre-requisite for development, as well as developed economies that will face more pressure to bridge the digital divide.
Against the backdrop of what needs to happen on investments, there is the risk that incumbents will double-down on the existing structure and lobby for more power which could be detrimental to new entrants bringing in technology and business disruptions.
[1] Following school closures, traffic increased by amounts ranging between 0% and 80% based on global samples. Average and median are about 33%. Speed decreased on average by 14% with a median of 9%. Following general closure, traffic increased on average by 15% and 10% median. The corresponding reduction in speed was on average 2% and median of 4%.
Here you can find examples of traffic reports from Verizon, AT&T, Vodafone, Sandvine, and Cloudflare.
In the course of our research, we got questions on the impact on 5G deployments, spectrum auctions and valuations, the consequences of virtualization and other topics. Contact me if such questions are of interest.