Cloudflare points to ‘rickety infrastructure’ following Verizon outagesCloudflare points to ‘rickety infrastructure’ following Verizon outages

Cloudflare, which describes itself as a connectivity cloud company, has released its Q4 Internet Disruptions report which looks into impacts and causes of internet outages around the world. It oversees roughly 25% of the world’s internet and operates in over 120 countries, we’re told, and so has a “bird’s eye view on how the Internet is functioning at a global level.”

There’s a lot of data points within it, but three incidents in particular are highlighted.

On November 12, some subscribers of Verizon’s Fios Internet service experienced disruption to their Internet connectivity, says the report, which referenced a post to the Outages mailing list which stated: “major multi-state Verizon Fios outage began at 12:28am EST, impacting Virginia, Washington DC, Maryland, and New Jersey, as well as parts of eastern Pennsylvania. Traffic from AS701, the autonomous system used by Verizon for their Fios service, dropped by approximately 30% around 00:30 Eastern time (05:30 UTC).”

“Verizon’s outage highlights rickety infrastructure” stated the accompanying email, and we’re told the outage caused traffic drops of 50-70% in some regions, affecting densely populated Northeast US states which underscores “the necessity for increased resilience in critical telecom infrastructure.” Verizon apparently never disclosed the root cause of the outage, which lasted roughly three hours.

Elsewhere the report observed a complete power outage in the island of Guadeloupe, with restoration taking over two days to complete.

The report cites an article published in The Guardian on October 25 which noted that “The French Caribbean island of Guadeloupe has been left entirely without power after striking workers seized control of the territory’s power station.” Workers entered the power station’s command room “and caused an emergency shutdown of all the engines”, according to the article.

The report says the power outage caused by this “emergency shutdown” resulted in traffic dropping nearly 70% as compared to the previous week, and recovery took significantly longer than expected.

The report also says that internet shutdowns in Mozambique were reported across multiple providers during election-related protests in October and November.

“On October 25 in Mozambique, mobile Internet connectivity across multiple providers was shut down after protests against the re-election of the ruling Frelimo party became violent. Starting around 13:00 local time (11:00 UTC), significant drops in traffic were observed across AS30619 (Telecomiuncacoes de Mocambique), AS37342 (Movitel), and AS37223 (Vodacom),” stated the report.  

Both Vodacom and Movitel experienced near complete outages almost immediately, while some traffic remained on Telecomiuncacoes de Mocambique until just before 02:00 local time (00:00 UTC) on October 26, we’re told.  

By way of conclusion, the report offers: “In addition to the outages and disruptions covered above, resilient Internet connectivity meant that two Baltic Sea cable cuts that occurred on November 17 and 18 had minimal impact. Whether accidental or sabotage, the security and resiliency of submarine cable infrastructure continues to be an important topic. The security and resilience of terrestrial cable infrastructure, as well as other critical Internet infrastructure, must also remain top of mind to help speed recovery from storms, earthquakes, military action, and power outages.”

The point of the report appears to be to provide a snapshot of the myriad types of disruption that faced networks around the world during the quarter and how quickly they were resolved, as opposed to provide a theme or wider point. You can read through the whole thing here.