Where to start with public cloud computing
April 4, 2011
The first step is to evaluate your workloads, says Mark White, CTO for Deloitte Consulting’s technology practice.
"That’s step one, and the conversation is really about the services catalog," he says. "If the company already has an IT services catalog, those entries represent various workloads. If not, you’d look literally at IT services less formally and their workloads."
For large enterprises, public cloud-appropriate SaaS workloads are those that classically have little automation, such as for marketing, or those that have had some automation, but little integration. CRM, customer care and human resource management systems fall in this latter category, he says.
"You’re looking for places where there are islands, so less legacy investment or less subsequent re-integration," White says.
"However, that doesn’t mean you can’t do the larger, more complex, legacy systems with integration. But our analysis shows the likelihood or timing for large enterprises to look at core systems such as integrated financials, manufacturing, inventory and supply chain management is much later in the public cloud cycle. And, in fact, you see that in the offerings from public cloud providers – few, if any, are looking at larger enterprises in those core spaces," he adds.
Once you’ve identified workloads that could work in the public cloud, the next step is to build the business case. This is no different from the norm. You’ve got to determine the investment, the return, expected performance characteristics, subscription terms, and evaluate the risks and obstacles – especially security and privacy sensitivities – as well as goals and enablers, White says.
Of course, if you’re planning a pilot-mode project, the business case needn’t be of Harvard Business Review caliber, he adds.
"One of the attractive things about public cloud is that it lends itself to a toe in the water. If it goes well, the likelihood that you’ll scale with a provider is high. If it doesn’t go well, you can easily recast and try again," White says.
Lastly, along with evaluating the business case, consider the internal stance. Knowing whether people and processes can support a cloud initiative is critical, he says.
"The same steps apply to IaaS and PaaS, as well, although these come from within the CIO shop so they’re about the business of IT and not the business of the business," White adds.
Here, he says, enterprise IT executives should think in terms of what Deloitte calls the "capacity cloud." Meaning, they’re running workloads in-house today for which the ability to grab capacity on demand would be beneficial or they’re interested in running a new workload but would like to isolate it outside the shop.
Of course, nothing about the cloud is quite as black and white. "There are obviously a lot of shades of gray but thinking along the lines of these characteristics is at least a good start," White says.