Skip to content
You are here: Home >> Blog >> Consumer Technology >> Numerous popular websites go down due to CDN error

Numerous popular websites go down due to CDN error

At about 11AM BST today (08 June), you may of had some trouble accessing sites like Reddit, Amazon, and the UK Government website. News sites – including those reporting on the outage – also went down. The Verge resorted to a Google Doc to bring live reporting (which was initially editable by the public but this was promptly resolved). Other websites affected include Stack Overflow and GitHub – both popular tools for developers, as well as PayPal and The New York Times.

What went wrong?

Initially, Internet users were left confused and expressed their frustrations using platforms that were still available, such as Twitter – which did stay online, but suffered a problem with emojis.

Similarly, I was able to load the Amazon website, but only in a plain-text form.

The below gallery shows examples of Reddit, gov.uk, The Independent, as well as Fastly’s own documentation being affected.

A collection of errors from multiple websites displayed to users around the world – Fastly’s own website was not spared.

Such wide-scale outages like this are fairly uncommon – and the disruption led to Breaking News splashes across the world both online and on TV, sparking debate on how the Internet relies on a small number of companies to provide key services.

Content Delivery Networks

Companies like Cloudflare and Fastly are CDN‘s. This means they provide edge servers that cache website content and files (including images) from origin servers which then get delivered to users from the nearest edge server to them. This often helps speed up many websites – including this one, which uses Cloudflare, as it avoids constant usage of the origin server. The problem is, there are not many CDN companies – so most turn to the big players in the market – and when they go down, many websites are affected.

The Fastly Outage

The CDN causing issues today was San-Francisco based Fastly, which describes itself as optimised for ‘speed, security, and scale’.

According to their status page, Fastly first reported issues at 09:58 UTC today, ‘investigating potential impact to performance with our CDN services.’ At 10:44 UTC, they reported a fix being implemented. The issue was considered formally resolved at at 12:41 UTC, stating:

Fastly has observed recovery of all services and has resolved this incident. Customers could continue to experience a period of increased origin load and lower Cache Hit Ratio (CHR).

https://status.fastly.com/incidents/vpk0ssybt3bj

At around this time, the affected websites began to recover, although some still experienced issues for some time after due to increased origin server load. However, reliable service has been restored for most now.

“We identified a service configuration that triggered disruption across our POPs (points of presence) globally and have disabled that configuration. A POP allows content to be sent from globally distributed servers that are close to the end user. “Our global network is coming back online.”

Statement from Fastly (via BBC)
Tags: