Aardvark Daily

Please visit the sponsor!

Someone broke the internet

When I fell out of bed this morning at 1:30am (pretty normal) I found the newswires saturated with reports that the internet was broken.

Well not the whole internet but certainly a significant portion that was reliant on Amazon's AWS cloud servers.

This resulted in crucial services, such as those operated by the UK government, a number of banks, popular sites such as SnapChat and a raft of other online properties being badly broken.

It's moments like these that you realise just how fragile the supposedly "fault tolerant" internet has become.

One of the very reasons the internet was built is in order to provide a fault-tolerant transport layer for data and that's pretty much the way it does work... until it doesn't.

Back in the day, the internet was made up of a highly distributed network of computers so the failure of any individual computer or the data connection to it wasn't really a big deal.

If website X on one computer went down the the websites at Y and Z on other computers at different locations, would carry on unaffected.

Unfortunately, because we're now becoming increasingly reliant on huge datacenters that create cloud-based services, the situation has changed quite significantly.

We no longer have lots of logically and geographically distributed computers serving up websites and other services. Instead, many of those have been replaced by virtual servers in those datacentres. This means that if something goes wrong at one of those datacentres or there's a failure of connectivity to that datacentre then huge swathes of the online world can simply disappear in the blink of an eye.

According to reports, the AWS outage last night had a world-wide impact, even affecting NZ services provided by Spark, TVNZ and Sky TV's streaming video services.

This event should be sending strong warning signals to companies and infrastructure providers alike. If a single point of failure such as this can have such widespread effects then surely that makes for a massive vulnerability that could be exploited by bad actors, for any number of reasons.

This also further highlights the vulnerablity surrounding the digital ID systems being rolled out by the UK and Australia. During the outage, UK government services were knocked offline and if digital ID had already been a mandatory thing in Britain the tens of millions of people may have been in a situation where they were denied access to essential services.

The single-point failure with massive outcomes also perhaps highlights the sage warnings in my last column where I pointed out the vulnerability of storing all our knowledge in the cloud.

Fortunately, at least to date, such widescale outages are rare events but I fully expect that they may become more commonplace in the face of growing cyberwarfare attacks from the state actors of nations such as North Korea, China and Russia. If they can effectively disable a huge chunk of a country's internet by taking down a single data-link or server farm imagine how attractive that would be to them?

AWS is not the only critical piece of infrastructure on which so much of the world's internet is reliant. There are also the root DNS servers and services such as CloudFlare that, if knocked offline, would bring the Net to its knees in a microsecond.

Carpe Diem folks!

Please visit the sponsor!

Here is a PERMANENT link to this column

Rank This Aardvark Page

Change Font