https://www.pexels.com/@deepanker70

The Guardian, eBay, Amazon, The New York Times to name but a few of the high profile websites that went down at the beginning of June. Panic set in as another global internet outage struck.

Just before 11am UK time, a company that few had heard of showed just how fragile our global internet infrastructure really is. #cyberattack started to trend on Twitter, speculation was rife, but the reason was relatively simple, someone misconfigured a server.

The little known company of Fastly is a huge content delivery network or better known when abbreviated to a ‘CDN’. Cloudflare is another example of such a network as is Amazon’s own version named ‘Amazon Cloudfront’. CDNs’ deliver content hosted in one place to users in another place. For example, a website hosted on servers in New York and users near New York will use those servers to quickly access that content. Whereas, users in the UK will find the website slightly slower if using those same servers in New York. This can be an issue, for video streaming, for example. The overriding issue is the speed of light, nothing can exceed the speed of light (that we presently know of) and given the distances involved the difference can amount to a few milliseconds even if the fibre-optic cable runs in a straight line, which it rarely does. That delay can slow things down, making it impossible to stream videos or even pay for a product online.

CDNs solve this via physical infrastructure called ‘edge’ servers, computers on the ‘edge’ of networks where the computation of data needs to take place geographically located closer to the user. They are the concept of processing and storing content data towards where that user is located. CDNs typically pick up the most used content and cache it closer to areas of dense population. ‘Mirroring’ is another method of hosting the same content on multiple servers located around the world but this is costly and few companies want to spend money implementing that.

The issue with the outage in this case was not a delay, but no data was being transferred …at all. A website being little slow is one thing, but for a popular website to be down is entirely another.

Over recent years, demand for fast loading and smooth access to websites has created the need for third party infrastructure which can take down a huge section of the internet if it goes down. The scope was huge in this instance, Reddit, StackOverflow, Github, Amazon, HMRC, eBay, Gov.uk and most news organisations went down. Fastly is mostly used by enterprise users such as the UK Government and when high profile websites have an issue, we notice in a very short space of time. The internet in it’s infancy was never designed to deliver huge amounts of data, so the need for storing that information on the ‘edge’ has been the recent big game changer.

99.9% of time, CDNs deliver content without any problems. Only when it goes wrong, we wake up and realise there is an issue. Concentration of infrastructure is done to make it more reliable and to do that is no bad thing, only it makes it more complex, costly and more likely to fail. Engineers may only think of technology whilst giving little thought to economy and we as users have all but lost patience with the early internet. It is a shame that engineers and developers work goes unnoticed apart from when a website or network goes wrong. Ancient Egyptians built the pyramids, Romans built roads, we hide the infrastructure away in anonymous boxes.

Thankfully, the issue was rectified within hours but should we take it as a cautionary tale of the fragile nature of our internet infrastructure? It is certainly another reminder that the internet is only as good as it’s parts. But given how much we depend on it, is the important lesson.

Further outages are to be expected, however, things could soon change as private companies have started to control more and more of the internet and what can be done on it, unease with that as been shown in both society and government. Will convenience be sacrificed to change the infrastructure that causes rare but mass outages is another thought provoking question.