Amazon Outage Highlights the Pitfalls of Cloud Hosting

March 27, 2017

Last month Amazon Web Services (AWS) experienced an outage. When their cloud-computing network went down, it brought with it a number of major Internet sites ― Quora, Business Insider, Netflix, Reddit and Slack ― and many more smaller sites. Observers reported that the event also impacted Mashable, a global, multi-platform media and entertainment company, prompting that company to resort to a tweet: "We can't publish our story about AWS being down because, well, AWS is down."

Amazon immediately worked on identifying and correcting the problem, but even outstanding recovery efforts still draw attention to the fact that there was a serious error.

One analyst looked past the single human error that directly caused the outage ― a typo in some pivotal AWS testing efforts ― to the other human errors that made the outage so dramatic. Oleg Dulin, writing for ComputerWorld, explains that many AWS customers didn't pay close enough attention to the difference between cloud "durability" and cloud "accessibility." Durability is your cloud provider's assurance that your data will not be lost, and most services have a phenomenal track record here. Accessibility, on the other hand, measures whether or not you can get to that cloud-hosted data; ratings here are significantly lower. Even with AWS's promise of 99.99% accessibility per year, there will be almost an hour where users will not be able to access their data.

It's also worth noting last month's outage was not a unique occurrence. While only Amazon knows the full details about AWS outages, DownDetector.com collects and publishes anecdotal evidence of problems with access. Since October 2015, they have identified 31 AWS outages of various significance.

To minimize the impact of unexpected cloud-hosting downtime, Amazon and other hosts use cross-region backups, cross-region replication, and smarter replication. However, as one analyst put it, "This incident will make enterprises think twice about moving certain workloads and apps to the public cloud and motivate them to look closely at the private cloud." In discussing more about the solutions to cloud outages, another writer explained that Big Cloud services have denied one of the primary features of the Web, exchanging resilient, de-centralized networks for convenient, although brittle, centralized systems. He recommends that your IT team look into distributing resources across multiple regions and also explore options for private cloud and hybrid cloud models.

Taking up the topic of hybrid models, Hostway, a smaller cloud-services provider, has argued that in hosted services arena, bigger isn't necessarily better. By leveraging the largest networks, Hostway says they are able to provide a better experience than any one of the largest clouds can offer. They offer a free white paper that elaborates on the details.

Accessibility requirements for accounting, CRM, and ERP software vary greatly from company to company. For some, downtime is just a slight inconvenience. For others, disrupted access for a single hour during the business day could cost tens of thousands of dollars. That's why aACE software can be installed on premises or hosted using FileMaker Cloud (which leverages Amazon) or set up with a customized architecture that is designed by your IT team.

Contact us today to talk about your company's business software needs.

Topics

Recent Posts