Cloudflare operates in more than 250 cities in over 100 countries, where we interconnect with over 10,000 network providers in order to provide a broad range of services to millions of customers. The breadth of both our network and our customer base provides us with a unique perspective on Internet resilience, enabling us to observe the impact of Internet disruptions. In many cases, these disruptions can be attributed to a physical event, while in other cases, they are due to an intentional government-directed shutdown. In this post, we review selected Internet disruptions observed by Cloudflare during the first quarter of 2022, supported by traffic graphs from Cloudflare Radar and other internal Cloudflare tools, and grouped by associated cause.
Internet outages caused by “earth movers” are more frequently caused by errant backhoes. However, two Internet disruptions in the first quarter were caused by more significant earth movement — a volcanic eruption and an earthquake.
The first impacted connectivity on the island nation of Tonga, when the Hunga Tonga–Hunga Ha'apai volcanic eruption damaged the submarine cable connecting Tonga to Fiji, resulting in a 38 day Internet outage. After the January 14 eruption, only minimal Internet traffic (via satellite services) was seen from Tonga. On February 22, Digicel announced that the main island was back online after initial submarine cable repairs were completed – the immediate return of traffic is clearly visible in the figure below. However, it was estimated that repairs to the domestic cable, connecting outlying islands, could take an additional six to nine months.
The second disruption, caused by a 7.3 magnitude earthquake off the coast of central Japan on March 16, was significantly shorter, and had a significantly smaller impact. The earthquake occurred around 1436 UTC, causing power outages that resulted in a loss of Internet connectivity in cities including Tokyo for several hours, as seen in the figure below. Almost exactly 11 years prior, a magnitude 8.9 earthquake also had a nominal impact on Internet connectivity in Japan, that time apparently due to damage to subsea cable systems.
Internet resilience is, of course, heavily dependent on the resilience of the underlying physical infrastructure, including data centers, terrestrial fiber, and submarine cables. Damage to this infrastructure often disrupts Internet connectivity.
Early in the morning of January 5, the Gambia was completely isolated from the global Internet. As evident in the figure below, the incident lasted over eight hours, between 0117 and 0945 UTC. According to a press release from GAMTEL, after the failure of the primary link (damage to the ACE submarine cable), traffic was routed onto two backup links through Senegal. However, these backup links also failed because they converged in a location that was ultimately identified as a single point of failure.
Around 2130 UTC on January 20, Internet traffic to Yemen dropped to near zero, as shown in the figure below, after ongoing airstrikes reportedly hit a telecommunications building in Al-Hudaydah where the FALCON undersea cable lands. The outage lasted four days, finally recovering around 2100 on January 24. The outage primarily affected YemenNet (Public Telecommunication Corporation), the state-owned telecommunications provider.
According to a published report, one of the cuts was on the Victorian (mainland) end, and the other on the Tasmanian side, with both cuts caused by “third parties”. A significant reduction in traffic between 0130 - 0800 UTC is visible in the figure below.
A reported Telecom Infrastructure Company (TIC) data center fire caused a four-hour Internet disruption in Iran on March 4. Telecom Infrastructure Company (TIC) is the monopoly provider of telecom infrastructure to all public and private operators in Iran. As the figure below shows, Internet traffic to the country dropped by approximately 20% at 0640 UTC, and recovered around 1030 UTC.
On March 15, ETECSA, the Cuban state telecommunications company, reported that a fiber optic cable had been cut on a public road in the capital that morning. The impact of this fiber cut on Internet traffic to Cuba and ETECSA is visible in the figures below, starting just after 1200 UTC, lasting for over six hours.
Although initially believed to be the result of a power outage (all too common in Venezuela), a March 24 Internet disruption in the country was ultimately due to a fiber cut. Internet traffic to CANTV customers in multiple Venezuelan states dropped significantly between 1140 and 1740 UTC, as seen in the figure below. In addition to this disruption, VE sin Filtro reported a number of additional multi-hour, multi-state Internet disruptions in Venezuela during the first quarter.
On March 31, Internet traffic to Telenor Pakistan dropped 60% between 0600-0745 UTC, as shown in the figure below. According to Telenor Pakistan responses to hundreds of customer complaints lodged via Twitter, the disruption was due to multiple fiber-optic cable cuts in several locations. Just after 1800 UTC, Telenor Pakistan Tweeted that services had been fully restored.
In addition to the physical infrastructure, reliable electrical power is also critical for resilient Internet connectivity. At a provider level, loss of power can take key data centers and routers offline, impacting connectivity for customers and other connected networks. Consumer power outages can take home/business routers and connected devices offline, forcing users onto mobile connectivity, assuming that is/remains available.
The interconnected electrical grids of Kazakhstan, Uzbekistan, and Kyrgyzstan all suffered outages on January 24 after Kazakhstan's North-South power line was disconnected due to "emergency imbalances". These power outages caused multi-hour Internet disruptions across all three countries starting around 0600 UTC, as the figures below show. The impact to traffic in Kazakhstan appeared to be fairly minor, while traffic declined significantly in Uzbekistan and Kyrgyzstan and took longer to recover.
A power outage across multiple counties and cities in Taiwan starting around 0100 UTC on March 3 caused a brief #Internet disruption. The figure below shows a nominal initial drop in traffic, though traffic remained lower throughout the next several hours. The power outage was reportedly caused by human negligence during annual repairs of a generator at the Hsinta power plant.
In addition to the fiber cut discussed above, Cuba’s Internet suffered a second disruption on March 24. A Tweet from ETECSA stated that a power failure had caused a disruption to voice service, SMS, and mobile data. Analysis of Internet traffic for both Cuba and ETECSA finds that the disruption started around 1230 UTC, and lasted for approximately 90 minutes, as shown in the figures below.
Although distributed denial of service (DDoS) attacks often target web or application servers in an attempt to knock a given website or application offline, such attacks that target network infrastructure can have more widespread impact, not only restricting access to sites and applications hosted within that network, but also disrupting connectivity for users attached to the network.
Such a DDoS attack targeted AS8867 (E-Gov - Tehila Project) in Israel on March 14. The figure below shows that Internet traffic to that ASN began to decline just before 1530 UTC. A published report notes that the websites of the interior, health, justice and welfare ministries, as well as that of the Prime Minister's office, were all taken offline as a result of the attack.
Unspecified technical causes
As discussed above, the underlying technical or physical causes of Internet disruptions are often easily identified, frequently thanks to social media or other communications from the impacted network providers. However, sometimes disruptions are observed that are correlated with a real-world (often political) event with no specified technical or physical cause, while other times disruptions are observed but are both uncorrelated and unattributed.
In Kazakhstan, an Internet disruption began on January 5 amid mass protests against sudden increases in energy prices. Starting around 1030 UTC, traffic from Kazakhstan dropped to near zero. The figure below shows that traffic returned to a regular diurnal pattern on January 11, but several apparent restorations of connectivity are also visible during the six-day disruption. These brief periods of connectivity appeared to align with televised speeches or announcements from the Zazakh president.
In Burkina Faso, heavy gunfire related to an army mutiny was reported early in the morning of January 23. A significant drop in traffic from the country was observed in Cloudflare Radar starting around 0915 UTC, with Orange, FasoNet, and Telecel Faso all seeing lower traffic volumes. As the figure below shows, the disruption lasted for nearly a day and a half, recovering around 2000 UTC on January 24.
Just after 2200 UTC on March 15, Yemen experienced a significant, albeit brief, Internet disruption, lasting just 30 minutes. As the figures below show, the disruption was primarily due to an issue at YemenNet. A published report claims that the disruption was due to a deliberate act by the Houthi coup militia.
Russian invasion of Ukraine
The Russian invasion of Ukraine has now been going on for over a month. In some cases, Internet connectivity has been collateral damage from the kinetic military action, while in other cases, targeted attacks on network providers and power outages have disrupted connectivity. Technicians from Ukrainian service providers have been risking their lives to keep the country online, and have been largely successful – Cloudflare Radar traffic data for Ukraine shows that as of the end of March, peak traffic levels are at 85-90% of pre-invasion peaks. An earlier blog post provides additional details about Internet traffic patterns observed in Ukraine during the first week after the conflict began.
Below we highlight just a few significant disruptions observed on major Ukrainian network providers in March.
Two brief outages were observed at Ukrtelecom during the second week of March, shown in the figure below. The first, on March 8, lasted for just over two hours, while the second one, on March 10, lasted for approximately 40 minutes. No root cause has been reported for these disruptions.
Later in the month, on March 28, Ukrtelecom experienced a ~15 hour outage, lasting from 0800 UTC to approximately 0100 on March 29, as seen in the figure below. A Twitter thread from the State Service of Special Communications and Information Protection of Ukraine explained that the outage was caused by “a powerful cyberattack” against Ukrtelecom’s infrastructure, and that “In order to preserve its network infrastructure and to continue providing services to Ukraine’s Armed Forces and other military formations as well as to the customers, Ukrtelecom has temporarily limited providing its services to the majority of private users and business-clients.” A LinkedIn post from Ukrtelecom also highlights the non-stop work that the company has been doing to re-establish telecommunications services in impacted regions across the country.
The figure below shows that around 2100 UTC on March 9, Ukrainian Internet service provider Triolan suffered a significant disruption, reportedly resulting from a cyber attack. Traffic began to gradually return after approximately 10 hours.
Despite occasional connectivity disruptions, the Internet remains remarkably resilient. This resiliency is increasingly critical as the Internet finds its way into more and more areas of everyday life around the world. In addition to providing a suite of solutions that support that resiliency, we use the data exhaust from these solutions to monitor Internet reliability, availability, security, and performance.