Points of failure
Last week many of the webs largest sites and services were having massive problems including Scotland’s own tech unicorn FanDuel, Googles Nest thermostats, and Amazons own Alexa devices. Web requests were either running slowly or not responding at all. As we move towards more devices being online in the Internet of Things (IoT) more single use devices will continue to be effected by outages such as these.
An issue at Amazons S3 US-EAST-1 datacentre was ultimately to blame, one of the most ironic casualties being their own status page which was set to load the red “failed” indicator from S3. This left the status page showing all “greens”.
Amazon have made a massive business out of hosting, few companies can compete with them in scale and their level of automation is almost second to none. But … by so many companies choosing to host with them, and with competition unable to keep up, do we risk putting too much of the web in one set of hands?
At GT4 we have confidently hosted with Rackspace for over a decade. They have their own datacentres, their own cloud hosting, their own automation APIs, everything a programmer could want. As a client, they offer one thing better than anyone else … a person to talk to.
Any time day or night I can lift the phone to Rackspace and get one of their technicians looking into any issues that we might have. Better than this though is that more often any issues we *may* have had we only find out about after Rackspace have identified and dealt with them
And as for that Amazon issue, turns out it was down to one guy running a single bad command. So if you think you’re having a bad day …