Christmas Eve left some folks at Amazon Web Services with coal in their stockings after Netflix, famously hosted on AWS, went black for much of Dec. 24, a problem the cloud services provider blamed on its load balancing infrastructure. While quickly resolved, the outage led to a flurry of complaints across social media sites, and highlighted the growing role cloud plays in the everyday lives of consumers. It also highlights the expectations of those consumers to have constant availability, an expectation we have taught them – perhaps unfortunately.
While the outage at AWS doesn’t really have much to do with the channel, the last part does – we’ve groomed end users to expect 100% availability and access. So what happens when the cloud doesn’t deliver? I think a quick search of social media and Netflix reveals that.
The lesson here is all about redundancy, a soap box I’ve been on before. If you build an on-premises network, you build in redundancy – backup, archives, failover systems, etc. But what about your cloud services offerings? Do you have a standard plan that pairs cloud with a redundancy plan? Because once your customers are accustomed to tapping into the corporate network anytime, from anywhere – you better be able to deliver.
This article also raises the question of who is really responsible for those redundant systems. While many AWS customers pay that vendor for redundant solutions available through Amazon, the media coverage of the outage acknowledges that the customer is also responsible, with Netflix “investigating” what happened as well as “revisiting its contingency plans.”
Let’s start the year with a new resolution: You won’t assume that your cloud services providers are fail proof, and you will uncover ways to support your customers when those clouds inevitability fail. You can thank me later.