Practical Analysis:How Much 'Cloud' Do You Really Want?

The many definitions of "cloud computing" are a conversation starter for some, a conversation ender for others. Many comments in our 2010 Cloud Computing Survey lamented that the term is still ill defined.

For those who view cloud computing as a conversation ender, it's time to get over it. Just as client-server computing was in the 1990s and Web-centric computing was for most of this decade, cloud computing, as ill defined as it may be, is real. Dismissing it because it seems like some other term might fit better is a career-limiting mind-set.

We also did this survey in 2009, and some trends are becoming apparent. First, the main proponents of cloud computing as a strategic means of delivering applications are still small businesses. These are the companies that bend their business processes to fit applications delivered as a service, and they're also the ones who don't currently have these apps in house. Big businesses use cloud services tactically, filling niches where they make sense.

Our survey also reveals that there's a lot more interest in software as a service and private cloud constructs than there is in platform and infrastructure as a service. That's not to imply that PaaS and IaaS won't succeed, but I do think interest in them is limited to certain markets and company types, whereas using at least some cloud ideas within private data centers is becoming mainstream thinking.

Whether or not you call it cloud computing, the ground is being laid by most of you for a more manageable and responsive data center. The vast majority of survey respondents have experience with server virtualization and, increasingly, storage virtualization. You're at least kicking the tires on networking consolidation and virtualization systems from Hewlett-Packard, Cisco, and others, and many of you are testing automation and orchestration tools.

For some companies, the goal is to replicate the services of a public cloud provider on a private level. That includes "self-service" allocation and deallocation of resources such as storage, network bandwidth, and virtual machines; support for multitenancy; chargebacks; and monitoring of service-level agreements. For others, the goals are more modest, mostly to automate data center functions so that IT teams can better meet demands and free up resources for other needs.

Cloud purists would say that approach is just data center automation and not "real" private cloud computing. However, IT pragmatists are likely to see that data center automation is where the big win is. That's because unless yours is a central IT organization providing services to divisional IT teams, you probably don't want to provide self-service, and your users wouldn't know what to do with an SLA. Do you really want Bob in accounting deciding when to add more virtual machines for SAP or Oracle Financials, or would you rather put some automation in place that helps predict what Bob needs?

It's also likely that the cloud purists looking to provide chargebacks and support multitenancy will need a different set of tools from new vendors. Those seeking automation will likely find that their existing vendors provide the products that meet their needs. The trick here is to not get caught up in the purity of the cloud effort, but to deliver what the company is likely to use, and then take your savings and move on to provide other business-changing and -enabling services.

Art Wittmann is director of InformationWeek Analytics, a portfolio of decision-support tools and analyst reports. You can write to him at [email protected].

To find out more about Art Wittmann, please visit his page.

More than 100 major reports will be released this year. Sign up or upgrade your InformationWeek Analytics membership.