Cloudflare's Q2 2024 Internet disruption roundup
Cloudflare's network spans more than 320 cities in over 120 countries, interconnecting with over 13,000 network providers. That breadth gives a unique vantage point on Internet resilience, and with Cloudflare Radar functionality released earlier this year, the impact of disruptions can be explored from both a routing perspective and a traffic perspective, at both a network and location level. This summary covers observed disruptions in the second quarter of 2024 and is not an exhaustive list.
As in previous years, nationwide exams in several MENA countries brought government-directed shutdowns. Both terrestrial and submarine cable cuts caused outages in numerous countries, with the ACE submarine cable a particular source of trouble. Maintenance, power outages, technical problems, and unknown causes also degraded connectivity, while Ukraine continued to suffer from Russian attacks.
Government-directed shutdowns
Exam-related shutdowns in Syria, Algeria, Iraq
Each spring, governments in several Middle East and North Africa countries order telecom providers to disrupt or shut down Internet connectivity for several hours per day over multi-week periods, aiming to prevent cheating on national exams. This pattern has been observed and covered in 2023, 2022, and 2021 across locations including Syria, Sudan, Algeria, and Iraq.
In June, we published Exam-ining recent Internet shutdowns in Syria, Iraq, and Algeria covering daily shutdowns in Iraq and Syria, plus two multi-hour daily disruptions in Algeria that appeared focused on content blocking rather than a full nationwide shutdown.
Iraq implemented a second round of shutdowns starting June 23, running through at least July 14. Some affected the same networks as the first round, while others hit networks in the autonomous Kurdistan region in the north. Among the latter, AS206206 (Kurdistan Net), AS59625 (Korek Telecom), AS48492 (IQ-Online), and AS21277 (Newroz Telecom) all implemented shutdowns on June 23, June 26, June 30, July 3, July 7, and July 10 from 06:00–08:00 local time (03:00–05:00 UTC).
Outside the autonomous Kurdistan region, networks including AS59588 (Zainas), AS199739 (Earthlink), AS203214 (HulumTele), AS51684 (Asiacell), and AS58322 (Halasat) implemented shutdowns between 06:00–08:00 local time (03:00–05:00 UTC) on June 23, June 24, June 26, June 27, June 29, June 30, July 1, and July 2. These followed the same approach as the first round.
Africa: Kenya and its neighbors
Ahead of planned protests against the Kenyan government's proposed "Finance Bill 2024" tax increases, a joint statement from multiple organizations urged the government not to enforce Internet shutdowns or information controls, warning of "disastrous economic effects." The Communications Authority of Kenya issued a press release stating it had "no intention whatsoever to shut down Internet traffic or interfere with the quality of connectivity."
As protests escalated on June 25, Kenyan Internet traffic dropped at 16:30 local time (13:30 UTC). The initial belief was that one or more undersea cables providing international connectivity had failed, a theory echoed in social media posts from Safaricom and Airtel.
Concurrent traffic drops appeared in Burundi, Uganda, Rwanda, and Tanzania. This is not unusual: submarine cable problems in one country can affect others that depend on it for upstream connectivity. A subsequently deleted post from MTN Uganda noted degraded Internet services "due to an outage caused by our connectivity supply through Kenya."
But the undersea cable explanation was quickly questioned within Africa's Internet infrastructure community. Kyle Spencer, Executive Director of the Uganda Internet eXchange Point, posted that the Kenyan government ordered sea cable landing stations to disconnect circuits. Ben Roberts, Group CTIO at Liquid Intelligent Technologies, stated "No cables are damaged this week." Outages on undersea cables also typically last days or weeks, not hours as this disruption did.
On June 26, Safaricom's CEO offered a revised explanation: the outage was "occasioned by reduced bandwidth on some cables." No further clarification came from Airtel or the Communications Authority. Some industry observers believe the disruption across the five countries was government-directed rather than cable-related.
Cable damage hits multiple regions
Digicel Haiti suffers repeated fiber damage
Digicel Haiti reported on April 28 that its international optical infrastructure in the Drouya area on National Road #1 had been damaged by cartridge impacts amid armed clashes in the area. The company stated that internet (data), SMS, MonCash, and international calling were affected, with all services restored to 100% at the time of the announcement. Traffic for AS27653 (Digicel Haiti) dropped from around 17:30 local time (21:30 UTC) on April 27 and recovered approximately 16:00 local time (20:00 UTC) on April 28, an outage lasting nearly 24 hours.
A further incident occurred on May 3, when the Director General of Digicel Haiti posted that the company suffered two more damages to its international fiber infrastructure at 2:00 AM. MonCash, SMS, and fiber optic connections were restored, with crews dispatched to address a landslide in the Canaan area. This disruption lasted approximately eight hours, between 02:15 and 10:30 local time (06:15 - 14:30 UTC), but had a nominal impact on traffic levels.
East African countries hit by submarine cable cuts
On May 12, issues with the EASSy and Seacom submarine cables again disrupted connectivity to East Africa, affecting countries that had experienced cable cuts nearly three months earlier. Traffic levels dropped just before 11:00 local time (08:00 UTC) across impacted countries. The initial impact varied: Kenya, Uganda, Madagascar, and Mozambique saw traffic drop by 10-25%, while Rwanda, Malawi, and Tanzania saw drops of one-third or more compared to the previous week. Tanzania, Madagascar, and Rwanda experienced the most significant overall impact.
Traffic returned to expected levels at different points during the following week, from May 13 in Kenya to May 19 in Rwanda. Repairs to the EASSy and Seacom cables were completed on May 31, while repairs to cables damaged in February were still ongoing as of July 9 due to their location in a war zone.
Chad's terrestrial dependency exposed again
A fiber optic cable cut in Cameroon disrupted connectivity for Moov Africa TChad customers on May 25. The three-hour outage, from 15:15 to 18:15 local time (14:15 - 17:15 UTC), was visible at the country level. The number of IPv4 /24 prefixes announced by Moov Africa Tchad fell from eight to three during the disruption. The event mirrored a January 10 incident where a fiber cut in Cameroon disrupted connectivity for over 12 hours, with significant volatility in announced IPv4 address space. As a landlocked country, Chad depends on terrestrial connections through neighboring countries, relying on limited cable paths through Cameroon and Sudan.
ACE cable outage spans multiple countries
A reported network interruption on the Africa Coast to Europe (ACE) submarine cable on June 5 disrupted traffic across networks in the Gambia, Mauritania, and Senegal. AS25250 (Gamtel), AS29544 (Mauritel), and AS37649 (Free/Tigo) all saw traffic drop around 23:00 local time (23:00 UTC). The outage lasted nearly 11 hours, with traffic recovering just after 10:00 local time on June 6. Mauritel experienced a near complete outage, while Gamtel and Free/Tigo appeared to shift traffic to backup links, resulting in less severe impacts.
Planned and unplanned maintenance
ACE cable repair work affects West Africa
Two months before the June 5 disruption, planned repair work on the ACE cable caused a different set of outages. A communiqué from the Ministry of Posts, Telecommunications and the Digital Economy in Guinea announced a planned outage on April 8, 2024, between midnight and 2:00 AM in Guinea, Senegal, Gambia, Sierra Leone, and Liberia, affecting Internet traffic and international calls. Traffic did not appear to fully return to expected levels after the two-hour repair window, and Senegal showed no observable impact despite being listed as an affected country.
GUILAB maintenance impacts Orange Guinea
Planned maintenance work on the ACE cable by GUILAB, the company managing Guinea's capacity on the cable, caused a multi-hour outage at AS37461 (Orange Guinea) and at the country level. The outage lasted from 12:15 to 15:45 local time (12:15 - 15:45 UTC). The maintenance was reported by Orange Guinea in two X posts that were subsequently deleted.
Power outages disrupt connectivity
Kenya grid disturbance
Kenya Power posted a "Power Outage Alert" on May 2 stating that a system disturbance on the grid at 5:40 PM (EAT) caused power supply disruption in most parts of the country. Internet traffic dropped sharply between 17:30 and 17:45 local time (14:30 - 14:45 UTC), lasting until approximately 21:30 local time (18:30 UTC), when Kenya Power posted a "Power Supply Restoration" notice. A longer-term view shows traffic peaks at the same time (22:00 local time) during the preceding two days, indicating the post-outage spike was consistent with normal evening usage patterns.
Ecuador nationwide blackout
A nationwide power outage in Ecuador on June 19 affected hospitals, homes, and the subway, causing a major disruption to Internet connectivity. Traffic dropped sharply just after 15:00 local time (20:00 UTC). Public Works Minister Roberto Luque posted that a failure in the transmission line caused a cascade disconnection, resulting in no energy service on a national scale. A later post noted that 95% of energy had been restored by 18:41 local time (23:41 UTC), with Internet traffic recovering to expected levels by that time.
Balkan countries hit by heat-related outage
A widespread power outage across Montenegro, Bosnia, and Albania on June 21 was caused by a sudden increase in power consumption due to high temperatures and heat-related impacts on electrical systems. The outage reportedly originated in Montenegro after a 400-kilowatt transmission line exploded. Power distribution systems in the region are linked as part of the Trans-Balkan Electricity Corridor, allowing the outage to spread across borders.
Electrical networks went down between 12:00 and 13:00 local time (10:00 - 11:00 UTC), with electricity suppliers beginning restoration by mid-afternoon and largely completing it by evening. Traffic from Albania, Bosnia, and Montenegro started dropping around 12:00 local time, reaching its lowest point at 12:30 local time in Albania and Bosnia and at 13:00 local time in Montenegro. Traffic returned to expected levels by 15:30 local time (13:30 UTC) as power was restored. Croatia was reportedly impacted as well, but no adverse traffic impact was visible at the country level during the timeframe other countries were affected.
Power Grid Attacks Hit Ukraine Connectivity
Russian air attacks on Ukraine's power infrastructure on May 21 caused Internet outages across multiple regions. The heaviest impact was measured in Sumy, where traffic dropped 82% below the previous week's levels at midnight local time (21:00 UTC). Kyiv, Kharkiv, and Vinnytsia also saw diminished traffic for several hours before returning to expected levels around 08:00 local time (05:00 UTC) on May 22.
Everyday Technical Failures
Not all disruptions stem from large-scale catastrophes. Routine technical issues can cripple access just as effectively. In Malaysia, Time Internet (AS9930) customers lost connectivity for nearly two hours on April 15 after both primary and secondary Secure DNS servers became unreachable at 17:38 local time. According to a Facebook post from the company, roughly 40% of customers were affected—the worst outage in the provider's history. Without access to DNS resolvers, browsers and applications could not resolve hostnames for any service, including those on Cloudflare's network. Traffic began recovering about an hour after the incident began, though the company never disclosed what caused both servers to fail.
Nepal's Upstream Dependency
Nepali Internet service providers rely heavily on Indian carrier Bharti Airtel for upstream connectivity—according to one published report, Nepali ISPs purchase 70% of their bandwidth from Airtel. On April 25, several providers including Vianet (AS45650) and Dishhome (AS139922) warned that service could be interrupted unless the government released foreign exchange to settle their outstanding payments, reportedly totaling USD $30 million. After Airtel gave notice on May 1, it cut off the delinquent providers the following day.
Vianet's traffic fell to near zero at 16:15 local time (10:30 UTC) on May 2 and took six hours to normalize. Dishhome, hit an hour later, saw significant degradation but not a complete outage. The difference likely stems from network architecture: Airtel is one of four upstreams for Dishhome's parent company, but only one of two feeding Vianet.
A second incident occurred on June 3 when routing issues on Airtel's network disrupted both Vianet and Worldlink (AS17501). Worldlink lost traffic between 12:15 and 14:00 local time (06:30–08:15 UTC); Vianet's loss spanned 12:15–13:15 local time (06:30–07:30 UTC).
Outages Without Explanations
Some providers acknowledge disruptions publicly, then never reveal the cause. CelcomDigi (AS10030) in Malaysia posted on X at 21:35 local time on May 21 announcing an outage and promising quick resolution. Twelve minutes later, a second post declared full restoration. But the traffic graphs show the actual disruption occurred much earlier, between 18:00 and 20:15 local time (10:00–12:15 UTC), with volumes returning to normal over an hour before the first social media post appeared.
Starlink's global subscriber base makes its outages unusually far-reaching. At 01:59 UTC on May 29, the satellite provider announced a network outage; 28 minutes later it reported the issue fully resolved. The event shows as a modest dip in traffic, followed by an intriguing spike after restoration—likely an automated connectivity check rather than withheld user demand, given how sharply traffic rose and then fell.
Government and Regional Cases
A near-total Internet outage struck Chad on June 5 between 08:15 and 12:00 local time (07:15–11:00 UTC). Announced IPv4 /24 blocks dropped 75% during the window. News coverage noted that only Starlink subscribers retained access, and pointed out that Chad has suffered recurring outages since 2016—sometimes from fiber problems, sometimes from government-directed shutdowns. The cause of this particular disruption remains unknown.
Reliance Jio (AS55836) in India, with a subscriber base exceeding 460 million, experienced two disruptions on June 18 between 13:15 and 17:15 local time (07:45–11:45 UTC). Each event lasted under an hour and cut traffic to roughly half of the prior week's levels, affecting both mobile and fiber customers. Jio has not offered an explanation.
Infrastructure's Fragile Layering
The pattern across these disruptions is clear: modern connectivity rests on a vulnerable stack of physical cables, technical systems, financial arrangements, and political choices. Redundancy improves resilience, but the most stubborn fault line is political. Government-directed shutdowns, such as those Chad has experienced, and payment disputes between providers in neighboring countries both highlight how non-technical factors can override well-engineered networks. Advocacy groups like AccessNow and its #KeepItOn campaign continue to press against these shutdowns, which carry real economic consequences for affected countries.
