Every outage has a silver lining
Amazon Web Services has apologised for a 14-hour outage on 20 October that affected its US-East-1 datacentre region in North Virginia, causing widespread global disruption to services including from Snapchat to Lloyds Bank Group.
The first I knew about the problem was when I had a day that didn’t feel like being slapped repeatedly in the face by an angry clown. A messaging application went down, you see. A messaging application that I use to communicate with one client (and only one, which itself illustrates how private platforms have fragmented communication and increased complexity for no good reason).
In my case it’s not clear if the AWS outage was actually to blame, either directly or due to some kind of knock-on effect, but the upshot for me was a happy one: a cancelled work meeting and a glorious absence of notifications that are a serious impediment to any kind of intellectual work.
The Internet, let’s not forget, grew out of a wacky and paranoid Cold War scheme and gained its first real users among researchers at universities. Yes, it has come a long way since then thanks to the development of things like content delivery networks (CDNs) and a glut of subsea cables laid during the dotcom boom, but the Internet’s commercialisation was laid on top of the network of networks, not built into it.
It would be an exaggeration to say that the Internet is held together using Coke cans and string, but not that much of one. At the very same time that businesses are trying to push everyone into digital channels, customers and staff alike, the infrastructure that underpins the Internet is not only increasingly centralised, it is composed of sophisticated systems running on often shonky core technology.
What went wrong
AWS’s largest US data centre region went down when its database service, network load balancers, and virtual server systems simultaneously malfunctioned, creating a cascade of failures that disrupted services globally from banking apps to doorbells. The latter would not be a problem if people would stop stapling WiFi chips to random household objects, though.
But AWS’s bad day at the database is far from the only way the Internet can bring global commerce and communication to an abrupt halt. DNS outages, too, are a major problem, for instance. DNS, or domain name system, servers translate Web addresses humans can remember (like, oh, let’s just say amazon.com) into the numerical IP addresses computers actually use.
When DNS fails, websites and services become unreachable even though they’re technically still running – it’s like knowing someone’s name but having no way to look up their number. A single misconfigured DNS record can render vast swathes of the Internet invisible.
Meanwhile, threats from ‘cyber criminals’ (really just criminals) and state actors alike are growing. Indeed, as geopolitical tensions continue to rise the two appear to be merging into a kind of nightmare gestalt. One or two snips of a few subsea cables and we’ll really find out just how ‘world wide’ the Web actually is when entire continents go dark.
In truth, we’ve drifted pretty far from the Internet’s distributed origins, building a house of cards on infrastructure that was supposed to be bomb-proof but turned out to be misconfiguration-vulnerable. Our solution? Double down. More centralisation, more cloud, more single points of failure. It’s like responding to a flood by building your house even closer to the river.
Still, looking on the bright side, the occasional outage might be the only thing that saves us from total digital immersion.
Subscribers 0
Fans 0
Followers 0
Followers