For the marketing types at cloud computing providers such as Amazon Web Services and Microsoft Azure, the holy grail is known as “five nines”—99.999 percent availability. It’s a term that means when you pay them big bucks to host your website, internet service, or to store data in their cloud, it is guaranteed, contractually, to be available 99.999 percent of the time.
This is known in the industry as “high availability.” An uptime of 99.999 percent equates to downtime of just 5.26 minutes a year. Customers, such as those affected by today’s Amazon S3 storage outage, sign contracts called Service Level Agreements (SLAs) which guarantee they’ll be receiving the highest availability possible.
Videos by VICE
The glaring issue here is that the guaranteed level is not 100 percent, and it never will be. Even Amazon’s S3 only offers four nines (99.99 percent) availability, and has to dish out free cloud computing credits to customers that don’t receive this level of service.
Time and time again it is proven that cloud computing outages can and will happen, forcing businesses offline and potentially causing their customers sizable chunks of lost revenue, and perhaps even starving their pets. Even today’s relatively minor outage meant that Motherboard was unable to upload images into our CMS, which allows us to publish our stories online.
Amazon’s S3 storage service effectively hosts images for thousands upon thousands of websites, including chat app Slack and yes, Vice. Today the issue is with its US-EAST-1 region, which is situated in Virginia. But Amazon’s cloud services have gone down before.
Indeed, even through three years of reporting industry news on cloud computing, not a single company executive would admit this to me, however. They’d harp on about redundancy regions (back up data centers, essentially), or evangelize about their bombproof servers. But it’s just not true: technical problems will always be a part of a technology, and as the world shifts more of its computing power online, rather than on a disk under a desk, the more even the most trivial of outages will cause major problems.
I guess the lesson here, like so many other industries, is not to put all of your eggs in one basket. Huge cloud players, led by Amazon Web Services, Microsoft Azure, and Google Cloud, are scrabbling over each other to host as many customers as possible—perhaps it’s best there will always be competition. And as we put more of our “things” in the Internet of (Hackable) Things, the costs of these outages continues to rise. It’s going from “the internet is down” to “oh shit, the internet is in all of my stuff, and now they’re down.”
