This post is presented in conjunction with The Internet Society.
February was a surprisingly quiet month for major Internet disruptions. In contrast to previous months, we observed few full outages or multi-day disruptions in the Oracle Internet Intelligence Map during the month. As always, there were a number of brief and unattributed disruptions observed over the course of the month, but the issues highlighted below were related to fiber cuts (and repairs) and likely problems with satellite connectivity. And while not yet a visible disruption, reports surfaced in February that Russian authorities and major Internet providers are planning to disconnect the country from the global Internet as part of a planned experiment.
Kicking off the month, Burkina Faso experienced brief partial disruptions to its Internet connectivity on February 1 & 2, as shown in the Country Statistics graphs below. The disruptions are also evident in the Traffic Shifts graphs below for AS25543 (Onatel), which is the country’s National Office of Telecommunications, holding a monopoly on fixed-line telecommunications there. Facebook posts from Onatel (February 1, 2) indicated that road work between the towns of Sabou and Boromo had resulted in a fiber cut, and subsequent posts made several hours later on both days reported that the resulting disruptions had been addressed.
In January’s post, we reported that a January 20th failure in the Tonga Cable had disrupted Internet connectivity to the island nation. On February 2, it was reported that full connectivity had been restored to Tonga. Since the country is GMT+13, the reports of restored connectivity align with the improvements seen in the Traffic Shifts graphs below for AS132579 (Tonga Cable Limited). During the period before the cable repairs were complete, Internet connectivity for Tonga had shifted to higher latency satellite links. As the graphs below illustrate, returning the Tonga Cable to service resulted in a significant decrease in latency as well as an increase in completed traceroutes to endpoints in this network.
On February 10 & 11, Internet connectivity in Zambia experienced several multi-hour disruptions, as shown in the Country Statistics graphs below. (Interestingly, we observed a marked increase in the DNS Query Rate metric as we saw drops in the Traceroute Completion Rate metric. While we would generally expect to see a correlated decrease in DNS request traffic during an Internet disruption, this increase may be due to applications and local resolvers making repeated resolution attempts. Due to the connectivity disruption, resolutions may fail to complete successfully, resulting in another attempt.)
Drilling down to a network level, we observed that networks downstream of AS30844 (Liquid Telecom), including AS7420 (Zamnet), AS36959 (AfriConnect Zambia), and AS37214 (Microlink Technologies) showed similar concurrent disruptions, as can be seen in the Traffic Shifts graphs below for those networks. Hai Zambia (a retail brand of Liquid Telecom) posted notices on both Facebook and Twitter, noting:
“Please be advised that we are currently experiencing a fault on our network which has resulted in Service degradation on all our services. Our engineers are working to rectify the problem and we would like to sincerely apologise for the inconvenience caused.
Closing out February, my colleague Doug Madory suggested that observed changes in Traffic Shifts graphs for networks in Vietnam may be related to a cable cut. As seen in the graphs below for AS45899 (VNPT) and AS7552 (Viettel), during the latter half of the day (GMT) on February 27, latencies to both network spiked alongside shifts in the relative percentages of upstream providers for traceroutes to endpoints within those networks. Further research by Doug found that HiNet (a service of Taiwanese provider Chunghwa Telecom) had posted a status update confirming damage to the Asia-Pacific Gateway (APG) cable at approximately 13:30 GMT, which corresponds with the events seen in the graphs.
During the latter half of the day (GMT) on February 18, the Central Pacific island republic of Kiribati experienced an Internet disruption, shown in the Country Statistics graphs below. Lasting from just before noon until midnight, it was likely caused by an issue with satellite connectivity provided by O3b, as the Traffic Shifts graphs below show a total loss of completed traceroutes through that path, along with increased latency, for the duration of the disruption. Currently reliant on satellite connectivity, Kiribati plans to connect to the Southern Cross NEXT submarine cable, which will connect the United States to Australia and New Zealand, when it goes live in 2021.
Just a few days later, Tuvalu experienced a disruption between 17:00 and 19:30 GMT on February 22. As the Traffic Statistics graphs below illustrate, traceroutes failed to reach targets on the island nation during the period of disruption, but the DNS Query Rate metric spiked during that same time. (See above for thoughts on why this may happen.) The Traffic Shifts graphs for AS23917 (Government of Tuvalu / Tuvalu Telecom) show that nearly all traceroutes reach that network through Asia Satellite Internet eXchange, and that a significant decrease in completed traceroutes was observed during the period of disruption. Tuvalu is not currently connected to any international submarine cables, although it is reportedly interested in connecting to the Tui Samoa cable.
On February 27, the British Indian Ocean Territory experienced several brief but significant disruptions to its Internet connectivity, shown in the Country Statistics graphs below. It appears that these disruptions are related to shifts in satellite connectivity providers for Sure Telecom, the sole commercial Internet connectivity provider in the territory. As the Traffic Shifts graphs below illustrate, traceroutes historically reached Sure Telecom through Cobbett Hill Earth Station Limited, but after a brief outage, started going through Intelsat instead. It isn’t known if this transition was due to a change in business relationships, a failure of Cobbett Hill’s service, or some other reason.
A proposed law introduced in the Russian Parliament in December 2018 mandates that Russian Internet providers should ensure that the “Russian Internet” will continue to function independently should access be cut off by other countries, and that all Russian Internet traffic can be routed through exchange points approved or managed by Roskomnazor (the “Federal Service for Supervision of Communications”).
According to a ZDNet article, Russian authorities and Internet service providers are planning an experiment to test whether these mandates are technically possible, and to gather insight and provide feedback and modifications to this proposed law. While no specific date for the experiment was cited, it is believed that it will take place before April 1, the deadline for submitting amendments to the proposed law.
If and when this isolation experiment takes place, we expect that it will be visible in the Country Statistics graphs for Russia in the Internet Intelligence Map. It is likely that the Traceroute Completion Ratio metric will fall to (near) zero for the duration of the test, and if the Russian networks withdraw from the global routing tables, then this will also be reflected in the BGP Routes metric. We will likely see a sharp drop, if not complete outage, in the DNS Query Rate metric, as DNS resolvers in Russia will be unable to contact our authoritative DNS servers. It is not clear whether routing traffic through approved/managed exchange points will be part of this experiment, or a separate one, and its impact within the Internet Intelligence Map will depend on whether these exchange points add any additional latency to traffic flows and/or interfere with traceroute or DNS traffic in any way.
In a November 2017 blog post (“The Migration of Political Internet Shutdowns”), we looked at the risk of Internet disconnection for countries around the world, based on the number of domestic network providers (autonomous systems) that have direct connections, visible in routing, to international (foreign) providers. (The 2017 post was itself an update to earlier 2014 and 2012 posts on the same topic.) Within this post, we classified Russia as “resistant” to Internet shutdown, because it had over 400 autonomous systems with connections to international networks. A review of Russia’s standing at the end of February indicates that they remain in the “resistant” category.
While it was encouraging to not see significant Internet disruptions in February caused by power outages, severe weather, DDoS attacks, or government direction, this is not likely to be a long-term trend. Power outages and severe weather are certain to impact Internet connectivity again in the future, and as DDoS attacks grow more powerful, it is likely that they will again disrupt connectivity at a national level. And while it is likely that we will see more nationwide government-directed Internet disruptions occur (like the exam-related outages seen in Iraq in the past), such activity is also becoming more surgical in nature, targeting selected sites and applications rather than taking the entire country offline, such as the blocking of Twitter, Soundcloud, Bing, YouTube, and Google experienced in Venezuela on February 27.