What the AWS Outage Means For Transitioning to Cloud Services

The Amazon Web Services (AWS) outage at the end of February – caused by a simple typo – had widespread effects across the web. Amazon S3 (Simple Storage Service) hosts a wide variety of data, and more than 140,000 unique domains make use of it.Amazon has identified human error as the cause. And although data wasn’t lost, there was a period of several hours where it was inaccessible. This translated to a chunk of time where consumers couldn’t make online purchases, stream entertainment, or view online content.

The failure of front-end operations was the most visible to users, but back-end data processing behind the scenes was also brought to a halt in many cases. According to IT analytics vendor Evolven, these human configuration mistakes can cost companies as much as $72,000 an hour in web app downtime.

Occasional and Inevitable

Server outages are inevitable, and no solution is 100 percent guaranteed to prevent them. In fact, Amazon’s S3 is still considered incredibly reliable, and it still easily meets its service-level agreement’s guaranteed uptime. And mistakes almost always provide learning experiences; engineers at Amazon have mitigated the potential for the same human error to occur in the future by limiting how quickly employees can remove server capacity.

Still, businesses can adopt public cloud solutions like AWS in ways that allow for contingency planning. Take Netflix, for example, which depends significantly on AWS – it didn’t suffer delivery problems on Feb. 28. Messaging service Slack also maintained basic functions (although users had problems with file sharing), and the company advised it had “degraded service.”

Instead of looking for nonexistent foolproof solutions, businesses need to be planning for failures and minimizing the fallout when they do occur. With that in mind, here are three precautionary measures any business can take to ensure advanced readiness concerning a server outage:

1. Adopt a Hybrid Approach

In this setup, data is split between the cloud and on-premise storage. Some businesses specify that various kinds of data not be allowed off-premise and only store less sensitive data in the cloud. In other circumstances, storing the same data in the cloud and on-premise provides redundancy and greater availability in case of disaster.

2. Work With Multiple Cloud Providers

It’s becoming more and more common for businesses to hedge their bets, choosing to rely on several cloud providers in case of outages. If a company has a fairly low tolerance for risk, this is likely a good option. This can also help prevent vendor lock-in.

3. Look at Availability Options, and Do a Cost-Benefit Analysis

There are diminishing returns when it comes to ensuring availability of data. Cloud usage can be designed for high availability (using strategies like zone-redundancy with a single provider), but it requires designing and testing that system for failover and recovery. This can ensure uptime during a failure, but it’s also costly. Thus, companies need to weigh how much uptime they actually require, as well as how costly downtime is for them.

While the number of sites affected by the AWS outage certainly attracted attention, the reality is that Amazon pretty quickly isolated the root of the problem and took the necessary corrective action to restore service. The fact that this limited amount of downtime did attract so much attention speaks to its rarity. Public cloud providers invest billions of dollars in infrastructure, and they’re much more reliable and secure than the average company’s on-premise data center.

That’s why the AWS incident shouldn’t (and likely won’t) cause a regression in cloud adoption. Instead of being wary of using public cloud space, prudent business leaders should recognize its strengths and protect themselves against its limitations. The cloud isn’t going anywhere. For business leaders and consumers alike, that’s a very good thing.

 

This article was written by Adam Levy of Magnet Solutions Group and appeared originally in Business.com.