2020 is by many means a year that everyone will remember. Every November in France, they celebrate the Beaujolais Nouveau, which will define the quality of wine for the year. While they don’t know the results of the 2020 vintage quite yet, more broadly for all of us, 2020 has clearly been a unique year.
Nobody was prepared for the pandemic. Overnight, companies had to adapt with virtually no visibility of a way out of the crisis. European authorities responded differently to the crisis causing each country to adopt various forms of shelter in place and timeframes. Each European enterprise adapted to the “new business as usual” where working from home has become the new norm and the Internet has become even more critical to businesses and individuals than ever. In 2020, ThousandEyes observed major outages across the European digital world that are worth talking about and will stay in the books for many years to come.
Top 5 Outages in EMEA in 2020
#1 May 11, 2020: Virgin Media – We are all heavily dependent on ISPs
Over the past several months during the COVID-19 pandemic, the increased dependence on the Internet and other third-party infrastructures has only compounded. Starting in March, when most of the lockdowns were put in place, enterprises had to adjust to work-from-home environments overnight. The Internet is made of thousands of ISPs that are interconnected, and when one of them is heavily impacted, it impacts the others—massive traffic changes eventually have a systemic impact that is longer lasting and bigger in scope. Virgin Media experienced widespread, intermittent outages of 15 minutes. While outages do often happen due to an operational issue or maintenance, what is interesting here is how long they lasted.
Figure 1: Example of an outage that lasted several hours within the Virgin Media infrastructure.
2020 clearly demonstrated our increased dependency on local ISPs that were experiencing massive shifts of traffic for which they were not always prepared.
#2 August 18, 2020: Equinix – You still need electricity to run a data center
A European-wide outage was observed on the 18th of August 2020, related to a power supply issue. Yes, we still require electricity to run data centers. This power outage impacted many users in the UK and ISPs leveraging Equinix services.
Equinix is a major Internet Exchange provider used by multiple content providers and ISPs due to its efficient network fabric on the Internet.
Figure 2: The impact of this outage that lasted over 30 minutes.
Figure 3: As we zoom-in on the outage, we observe multiple interfaces being impacted and all of them are located in the UK.
Figure 4: Equinix UK updated the community with the details of the current outage on Twitter.
This outage reminds us that we still require electricity to run the Internet. It also highlights our reliance on hardware, in this case electrical: without electricity there is no software.
#3 July 17, 2020: Cloudflare – The impact of shutting the front door service
Many organizations rely on Cloudflare to protect their services from DDoS attacks. Have you ever seen a CAPTCHA before connecting to a website? Well, that’s most probably delivered through Cloudflare as a front door to a web app through a DNS service.
Maintenance is fundamental in upgrading capacity, optimizing infrastructure and implementing routine tasks. However, the 17th of July 2020, there was a large outage when an error in configuration meant all DNS requests were sent to a single Cloudflare location, causing a 30-minute outage for the following locations in Europe:
Figure 5: User locations experiencing outages services served by Cloudflare service.
This outage demonstrates that DNS is a key service that needs to scale and is often assumed to be always working. This outage clearly reinforces the importance of the DNS service.
#4 August 30, 2020: CenturyLink – Almost 4% of the Internet went down on this day
30th of August 2020, is a historical date for the Internet. Due to a maintenance action, Level 3/CenturyLink, a major transit service provider, caused outages for almost 4% of the Internet during 5 hours. The outage was complex as even though customers were trying to mitigate the outage by rerouting traffic through other service providers, Level 3/CenturyLink continued to announce routes through its network causing a blackhole.
Figure 6: Level 3 continues to advertise stale routes despite services withdrawing routes.
This important outage shows that major ISPs can have a systematic impact of traffic patterns on the Internet.
#5 September 28, 2020: Microsoft 365 backend disruptions
Microsoft 365 experienced a backend disruption at the end of September causing users not be able to log in the Microsoft 365 service. The outage lasted at least 2 hours on the 28th of September from 21:20 PM UTC to 23:50 PM UTC.
Figure 7: Diving into the detail of the disruption, we observed the issue related to the authentication process into Microsoft 365 service.
What we learned from this incident is that third party dependencies are becoming even greater as we adopt more services from cloud providers.
2020 has been full of events on the Internet ranging from large traffic shifts requiring ISPs to re-engineer their network to cloud providers running into operational errors causing large scale outages. Clearly, 2020 has taught us that gaining intelligence into the Internet will help companies stay resilient. If you are ready for a fresh start in the new year, ThousandEyes will give the opportunity to gain unparalleled visibility and insights into digital experiences from every cloud to every employee and customer. As we begin 2021, our solutions can help you with increased visibility to resolve incidents quickly—schedule your demo today.