On July 3, Facebook suffered a glitch that left users unable to load images, videos and other data across its apps, including Facebook, WhatsApp, and Instagram. Users were unable to load photos in the Facebook News Feed, view stories on Instagram, and download images and videos on WhatsApp. Facebook’s Messenger service was also affected. It took the company about eight hours to resolve the issue. On DownDetector, a website that tracks outages, live outage maps for these services showed that the US, Western Europe, India, Malaysia, Indonesia, Japan and parts of Australia were the most affected. MediaNama’s journalists also noticed these problems last night.

Screenshot of WhatsApp showing the error message

MediaNama journalists were unable to download images on WhatsApp on July 3.

According to Down Detector, users started reporting issues around 7 pm IST (9:30 am EDT) on July 3. At its peak, DownDetector received 7,656 reports related to Facebook, 14,753 related to Instagram, and 1,630 related to WhatsApp. Since Facebook and Instagram were equally affected, users resorted to Twitter, and #Facebookdown and #instagramdown started trending globally. However, direct messaging on Twitter was also affected.

What did Facebook say?

Driving the irony home in 280 characters, Facebook tweeted as the outage started:

Eight hours later, at 2:36 am IST today, Facebook tweeted that the issue had been resolved, and said that it had been triggered during “routine maintenance operations”.

Instagram similarly tweeted as the outage began, “We’re aware that some people are having trouble updating or sending images and videos on Instagram. We’re sorry for the trouble and are working to get things back to normal as quickly as possible. #instagramdown.” The company later sent out a tweet announcing that the issue had been resolved: “We’re back! The issue has been resolved and we should be back at 100% for everyone. We’re sorry for any inconvenience.”

Website displayed alt-text instead of images

The Verge reported that during the outage, as the platforms suffered from image display issues, the website displayed alternative text instead of the images. Alternative text, or alt-text, is used for accessibility reasons. The descriptions and tags in alt-text are used to describe photos and videos to users with visual impairments. Alt-text can be entered manually, but Facebook has been using machine learning to automate the process by predicting what is in the picture since at least April 2016. The Verge clarified that it was not clear if these tags were used for ad-tech purposes as well.

Facebook and the centralisation of online communication

The latest outage brings into stark relief the fact that three of the biggest communication platforms in the world are owned and operated by one company. This, coupled with the New York Times report that Mark Zuckerberg is planning to integrate WhatsApp, Instagram and Facebook Messenger to enable interoperability, raises serious concerns about the centralisation of communication services. Facebook has more than 2.3 billion monthly users through its three services. An outage like yesterday’s, as a result, affects a third of the world’s population, and there’s no mechanism for redressal. Given that these services are used not only for private communication, but also power businesses, a technical failure at this scale translates into tangibles losses not just for Facebook but for the numerous companies, small and big, that rely on its various platforms.

Are outages becoming more common?

  • On July 2, Google Cloud Networking suffered from a load balancing issue which affected users on the east coast of the US. It took Google more than 24 hours to resolve the issue.
  • On June 24, Verizon, an American internet service provider, caused a major BGP misroute and consequently started routing most of the internet traffic though a small company. As a result, many Internet services such as Amazon and Cloudflare were affected. Cloudflare compared this outage to a ‘heart attack’. Ironically, the aforementioned DownDetector went down because it is hosted on CloudFlare.
  • On June 18, Google Calendar was down for more than three hours across the world.
  • A month ago, on June 4, Google Cloud Platform suffered major outages in the US which knocked out YouTube, SnapChat, Gmail, Nest, Discord and a number of other services.
  • On March 13, Facebook, Instagram, and WhatsApp, all went down to some degree globally in the company’s worst outage to date. Problems persisted for more than 24 hours and the company attributed the downtime to a ‘server configuration change’. Services such as Spotify and Tinder, which use Facebook credentials, were unable to authorise logins.