Bart Perkins: Cloudbursts ahead
Moving to the cloud may reduce infrastructure costs and headaches, but clouds have their shortcomings. When they rain, millions can quickly become drenched. In the past year, Amazon, Microsoft, Google and other providers experienced problems, from minor disruptions to major outages. A June 2012 headline captured the fallout: "Modern life halted as Netflix, Pinterest, Instagram go down."
The service interruptions experienced by those companies and others disappointed countless consumers, but it was worse than a disappointment for the businesses themselves. They had come to depend heavily on cloud reliability; when the cloud services they had put their trust in failed, it was as if they had ceased to exist. All the outages were temporary, of course, but revenues were lost during the downtime, and afterwards customers wrote blog posts expressing everything from disappointment to anger, with some proclaiming that they would take their business elsewhere.
Organizations that depend on cloud services need to manage four areas to help ensure that their dependence isn't a liability:
• Providers. Ideally, a cloud provider should be managed, monitored and measured like other critical IT suppliers. It is naive to take an "out of sight, out of mind" approach with cloud providers. Begin by setting clear performance goals with well-defined metrics. Assign staff to monitor performance and manage the supplier relationship.
• Resiliency. All cloud providers suffer periodic service degradation and occasional full outages. Unfortunately, restoration of normal service may not be fast enough to meet your business needs. On its blog, Netflix says, "It is still early days for cloud innovation, and there is certainly more to do in terms of building resiliency in the cloud." The blog describes the steps Netflix is taking to improve resiliency within Amazon's cloud.
Your enterprise architecture must also be designed for resiliency. Typical approaches involve spreading business functions, data and other assets across the cloud world. Most enterprises separate entities geographically within a single provider's cloud. The truly paranoid may want to spread their assets across multiple cloud providers.
• Executive expectations. IT professionals tend to be realistic about the reliability that is possible in the cloud, but a lot of executives on the business side expect well-known cloud providers to offer flawless service. They want "dial-tone" reliability, like the service provided by the old Bell system. But that's a standard that cloud suppliers can't meet now and are unlikely to fulfill anytime soon. (And even Ma Bell experienced occasional service problems.) And even if cloud vendors do one day offer dial-tone reliability, it would likely carry a premium price tag and would be cost-effective only for high-end products.
• Customer relations. Customers may become angry when their favorite services or phone apps are slow or temporarily unavailable. Your customer service staff's response must be sympathetic, informative and timely. When outages occur or service levels are significantly hampered, acknowledge the issue, apologize profusely, post status updates regularly, and share preventative measures as they are developed.
While the cloud offers a valuable alternative to extensive and expensive infrastructure, it's not yet perfect. When IT comes under attack for flawed cloud service, remind executives what prompted the move to the cloud: the difference between costs incurred with a cloud provider and those required to build and operate an enterprise infrastructure. Most good bargains require some level of trade-off, and clouds are no exception.
Bart Perkins is managing partner at Louisville, Ky.-based Leverage Partners, which helps organizations invest well in IT. Contact him at BartPerkins@LeveragePartners.com.
Read more about cloud computing in Computerworld's Cloud Computing Topic Center.