The dns outage at Akamai that made large numbers of sites worldwide inaccessible on Thursday evening was due to a bug caused by a software configuration update. The outage lasted for an hour.
The configuration update caused a bug in Akamai’s DNS system at 5:46 PM on Thursday. Rolling back this update resolved the issue an hour later. Which reports Akamai in a short update. According to the company, the outage had “impact on some customer websites.” According to the company, no attack took place on Akamai’s platform. The company does not provide further details about the update and the impact.
On Thursday evening, a large number of websites worldwide were inaccessible due to the DNS failure. Many companies purchase Akamais Edge DNS service, precisely to keep websites available, even if they are the target of DDoS attacks, for example. Akamai’s IP Anycast-based service uses thousands of name servers and more than 140 cities worldwide to resist attacks.
That it is a risk that sites rely so much on services from large parties to be accessible, was apparent earlier this year when the CDN service Fastly was disrupted. This outage also resulted in sites being inaccessible worldwide. Due to the failure at Akamai, sites of media and banks were inaccessible, among other things. The size of the outage appeared to be larger than it really was on Downdetector: the site uses social media postings to provide an indication of outages. Many users posted reports of potential outages to Google, AWS, and Cloudflare, but these companies have their own DNS services that were not affected.
Network analysis agency Thousand Eyes has mapped out the impact of the outage. According to the company, the impact varied widely from customer to customer. Some of the customers in Akamai’s CDN network were not reachable at all, while Amazon was, for example, because it uses other CDN networks and its own DNS service in addition to Akamai.