How a Google Cloud outage brought down the (smart) house

google cloud outage smart home
Image credit: Dale A Stork / Shutterstock.com

You might have noticed last week that Google Cloud experienced a serious service outage that affected a number of its services. But it did more than that – it also illustrated what happens to smart homes and other IoT apps that depend on a single cloud provider.

In short, when that provider goes down, so do your smart appliances. Or at least your ability to control them remotely.

On June 2,Google started what was supposed to be some routine maintenance for a few servers on its Google Cloud Platform, which requires rerouting the jobs those servers were handling elsewhere. Two misconfigurations and a software bug later, Google’s automation software inadvertently started descheduling jobs in multiple locations at once, which resulted in network congestion so severe that Google’s access to its own management tools to fix the problem was also mired in gridlock, reports Wired.

The network traffic jam knocked out Google services such as YouTube and G Suite (which includes Gmail, Docs and Calendar) for several hours, as well as third-party apps on GCP like Shopify, Snapchat and Apple’s iCloud.

Notably, the outage also affected Google’s Nest smart home products. According to Fast Company, numerous Nest customers tweeted that their Nest thermostats, smart locks, and cameras were inaccessible during the downtime. Sample tweet:

Important clarification: this is not to say that Nest-connected appliances didn’t work at all. It was still possible to operate them manually. What was down was the networked remote control functions that run on GCP.

Obviously, this matters because one of the selling points of smart homes is being able to control your connected appliances and locks and such when you’re not home.

If you were at work, for example, and wanted to adjust the thermostat so your kids or pets are comfortable, or check the home security cam, or unlock the door to let a friend in, you wouldn’t be able to do any of those things if the cloud is down (or at least the part of the cloud hosting your app).

These sound like minor inconveniences, but that’s because connecting home appliances to the internet is still a new enough thing that’s optional – the appliances don’t need the internet to do what they were designed to do (cool or heat a home, lock a door, make coffee, etc). The Google Cloud outage gives us a glimpse of what could happen to IoT devices that do depend on internet connectivity and the cloud to function properly, such as drones or self-driving taxis. Such devices are hopefully being designed with a back-up plan in case either the internet connection or the cloud suddenly becomes unavailable, rather than simply assuming one or both will always be available.

The irony is that the internet is supposed to be distributed and redundant. But as Professor Zeynep Tufekci observed on Twitter, with companies like Google Cloud and Amazon Web services dominating the public cloud, the result has been a concentration of resources that can become single points of failure, whether it’s due to an internal error or an outside attack:

If nothing else, this sounds like a good argument for multi-cloud services, which is generally pitched as a way to avoid vendor lock-in, but could also potentially mitigate the impact of outages. That said, I’m not sure how Google would feel about distributing its services on other clouds when having control of the servers and data centers the services run on is why they built their own cloud in the first place. (The same goes for AWS and Facebook.)

Whatever the solution, if we’re serious about pursuing an all-digital world where everything is connected, this is something we probably maybe should be thinking about looking at.

Likes
Please feel free to share

Be the first to comment

What do you think?

This site uses Akismet to reduce spam. Learn how your comment data is processed.