We need to rethink the resilience of the global Internet infrastructure

By on 11 Jan 2022

Category: Tech matters

Tags: , ,

2 Comments

Blog home

The Internet is a resilient distributed system that has successfully withstood several natural and human-made disasters in the past. However, today, Internet researchers and operators overlook a low-probability, high-impact event that poses a significant threat to the Internet: solar superstorms.

Coronal Mass Ejections (CME), popularly known as solar storms, are a directional ejection of a large mass of highly magnetized particles from the sun. When the Earth is in the direct path of a CME, these solar particles will interact with the Earth’s magnetic field and induce electric fields on the Earth’s surface. This can cause Geomagnetically Induced Current (GIC) to flow through long conductors with ground connections located far apart on the Earth, such as power cables and fibre cables with repeaters.

Fortunately, not all CMEs are powerful enough to cause substantial damage. The estimates of the probability of a solar superstorm causing massive outages range from 1.6% to 12% per decade. However, the risk is not uniform across years. The sun goes through cycles of approximately 11 years, and CMEs are more likely to occur during solar maxima. Moreover, the strength of solar cycles also varies with a period of roughly 100 years. By chance, modern technological advancement coincided with a period of weak solar activity. The sun is expected to become more active in the near future. However, we have a limited understanding of whether the current infrastructure is resilient against powerful CMEs since it has not been stress-tested under peak solar activity.

The largest solar events on record occurred in 1859 (the Carrington event) and 1921. They triggered extensive power outages and damaged the communication network of the day, the telegraph network. If a storm of a similar scale occurs today, Internet services could be affected in two ways: by direct damage to Internet infrastructure, or indirectly, through power outages. While the impact on power grids is better understood, we have a limited understanding of the impact on Internet infrastructure.

What infrastructure is at risk under solar storms?

The most vulnerable components of the Internet’s infrastructure are satellites and long-distance fibre cables. Localized infrastructure such as data centres can be protected from CME-induced electric surges using relatively inexpensive Transient Voltage Surge Suppressors (TVSS).

Threats to satellites, which are directly exposed to solar particles, include damage to electronic components and extra drag on the satellite, particularly in Low Earth Orbit (LEO) systems such as Starlink, that can cause orbital decay and uncontrolled re-entry to Earth.

Read: Everything you wanted to know about LEO satellites, part 1: The basics

Long-distance fibre cables are susceptible to damages from GIC due to the presence of the long conductor along the length of the cable used to power the repeaters.

Due to the orientation of the Earth’s magnetic field, higher latitudes are at a significantly higher risk. An analysis of the distribution of Internet infrastructure components — submarine cable endpoints, data centres, Internet Exchange Points (IXPs), DNS servers, and so forth — shows a higher concentration of these elements at higher latitudes.

Moreover, the investigation of submarine cable connectivity properties reveals that the risk is not uniform across various locations. For example, while the submarine cables connecting the US to Asia are more evenly distributed across the West Coast, the cables between the US and Europe are concentrated in the North East of the US. Thus the US is more vulnerable to disconnection from Europe. In contrast, Asia has relatively higher resilience, with Singapore acting as a hub with shorter connections to several economies and the presence of almost all cables restricted to lower latitudes.

Rethinking the resilience of the global Internet infrastructure

The first step towards quantifying the risk due to solar storms involves developing failure models for submarine cables. GIC models designed for power grids cannot be directly applied to submarine cables for several reasons: the large distance of separation between the ground points subjecting the cable to multiple electrical fields (ground points separated by thousands of kilometres in submarine cables vs hundreds of kilometres in power grids), presence of the ocean, and the coastal effect that amplifies induced electrical fields near the coast where submarine cable ground connections are located. There are ongoing efforts to model the vulnerability of submarine cables by geophysicists.

Even if the risks to submarine cables are minimized, the Internet is vulnerable to indirect outages due to power grid failures. Hence, we need to rethink the resilience of the global Internet infrastructure and applications that rely on it. This can be achieved in many ways:

  • Improving the resiliency of the topology by adding cables in less vulnerable regions.
  • Balancing the distribution of infrastructure components across latitudes.
  • Devising strategies to prepare for an impact (a CME that originates in the sun reaches the Earth only 13 hours to 5 days later).
  • Designing solutions for piecing together a partitioned Internet.
  • Expanding test scenarios to include large-scale failures and network partitioning.

To learn more, check out my recent SIGCOMM paper or watch the talk below.

Sangeetha Abdu Jyothi is an Assistant Professor in the Department of Computer Science at University of California, Irvine and an Affiliate Researcher at VMware Research.

Rate this article

The views expressed by the authors of this blog are their own and do not necessarily reflect the views of APNIC. Please note a Code of Conduct applies to this blog.

2 Comments

  1. Sandeep Goswami

    It is not just the internet infrastructure but the electrical infrastructure that needs to be robust during CME. The Internet can not work without power. This is where a system needs to be put in place.

    Reply

Leave a Reply

Your email address will not be published. Required fields are marked *

Top