A slew of websites operated by financial institutions, governments and airlines including Hong Kong Exchanges & Clearing and Australia’s central bank went down briefly Thursday in the second global internet outage in two weeks.
Some of the outages, including those that affected Commonwealth Bank of Australia, Westpac Banking and Australia & New Zealand Banking Group, were linked to a failure at Akamai Technologies, which helps clients manage web services, people familiar with the matter said, asking not to be identified discussing internal affairs. The Reserve Bank of Australia was forced to cancel a scheduled bond-buying operation Thursday, blaming “technical difficulties.” The central bank said it had adopted workarounds and its website was now back up and running.
The widespread downtime recalled an hourlong global outage earlier this month, triggered by a software failure at the content delivery platform Fastly. The resultant cascading failures, which affected services from Amazon.com to Shopify and Stripe, served as a stark reminder of how exposed the world’s biggest websites are to the impact of disruptions ranging from simple human error to coordinated cyberattack.
Akamai said in a statement it was aware of the issue and “actively working to restore services as soon as possible.” The website tracker Downdetector.com initially flagged hundreds of user complaints about outages affecting Southwest Airlines, Delta Air Lines and Automatic Data Processing. Other websites pinpointed included those operated by Vanguard, E-Trade and Navy Federal Credit Union.
Many of the websites affected on Thursday recovered within the hour, some after rerouting to other providers. Companies including Hong Kong’s exchange and Southwest said they were investigating the incident, without elaborating. “The pause in connectivity did not impact our operation,” Southwest said in an emailed response to questions.
It was unclear what triggered the incidents Thursday. In Fastly’s case, a valid software configuration change by one of its customers triggered a previously undiscovered bug, introduced during a May 12 software deployment. Fastly quickly identified an issue with its content delivery network and announced it was rolling out a fix just 46 minutes after acknowledging a problem. Sites began to spring back to life soon afterward.
Akamai is one of a number of high-level website and application hosting services that large enterprises use to serve content to millions of users simultaneously.
Rather than hosting all website content on a single set of servers in one location, Fastly’s so-called edge computing model puts servers in dozens of locations, allowing websites to serve pages to users from physical locations closest to them. This cuts lag time, speeding up page-loading and spreading the burden on individual servers.
These vast and complex setups are run by just a few companies, such as Fastly and Cloudflare. The global edge computing market was valued at $4.68 billion in 2020 and is expected to expand at a compound annual growth rate of 38.4% from 2021 to 2028, according to a recent analysis by Grand View Research.
While these setups usually work perfectly, their complexity means that even a simple error in a configuration file can trigger chain reactions of outages. For users, most of whom rarely need to think about how the internet works, that can come as a shock.
