If you were ranking the most talked-about topics of 2011, Cloud Computing would probably fall somewhere in between Charlie Sheen and the Royal Wedding. But, in most cases, that talk hasn't translated into a groundswell of action. According to a recent Alsbridge eSeminar survey, one of the stumbling blocks appears to be the initial planning – namely creating a viable Cloud strategy. How do you start the process? What factors do you need to consider? How can you develop an approach that best mitigates risk?
We asked some industry experts to weigh in with their advice.
The First Step: Change Your Mindset
"To start with, I think it's important to get past the hype and see Cloud for what it really is: a new computing model that drives efficiencies," explained Chris Mankle, chief technology officer for ACS, A Xerox Company. "Companies used to own their own telephone systems. Now, they get these services from a Telco in a utility-based model. Cloud delivers infrastructure services on a utility model, just like you buy phone service or kilowatt hours of power."
While the genius of Cloud is the flexibility to add or subtract capacity on the fly and the opportunity to turned fixed costs into variables, it is not the end-all and be-all. The greatest benefits come from using this model where it makes the most sense for that company. A start-up will use the model differently than an established enterprise.
"It's also important to note that there are different solutions within various cloud models. Public, private and virtual private clouds all provide different options," said Ryan Reed, product marketing manager, Cloud Computing for HP Enterprise Services. "Your ultimate solution will not be 'the Cloud, but a hybrid of traditional IT and specific types of Cloud with the ability to manage across both environments."
Identify The Business Problem
Although most people focus on the technology itself, the foundation of your Cloud strategy should be all business – namely, what do you want to accomplish?
"Everything starts with identifying the business problem you want to solve with the Cloud model," said David Mitchell, director of enterprise operations for Sabre Holdings. "Do I want to incorporate automated provisioning to solve my delivery problems? Do I want to throw hardware at a new release that's not working to increase my technical efficiency? Identify the business problem then see if that problem can be solved by Cloud. Otherwise, you're just going after new technology for technology's sake."
Inventory What You Have Today
Before you can determine your options in Cloud, you need a detailed inventory of your current infrastructure.
"List out what you have today to deliver business functions and the scope of each system in those key areas," Mankle said. "How is each system used by employees or customers; what does each system talk to – and is the reach global or regional?"
In addition to the technical overview, companies should identify any constraints, like older software releases that need updating. Also make sure the base requirements of your infrastructure will fit into a Cloud environment. Unique hardware platforms or vertically scaled systems simply don't translate in this model.
Identify Apps That Could Most Easily Cross Over
While understanding the environment is important, the true "make or break" of what you do in Cloud are the applications. These are the undisputed rock stars of the IT infrastructure and require thorough review upfront.
"Make sure you look at the security, storage growth rate and compliance issues for each application," Mankle said. "Review how you test your environment and whether or not these steps hold up in Cloud. Look at how you do audits or deal with failures in a traditional environment. Also document your inbound and outbound networking requirements for each app, because these could impact costs."
Although the optimal Cloud entry point will vary from company to company, experts agree on one thing: less critical, low-demand applications are typically the best place to start.
"I think you go for the two extremes when considering a Cloud strategy – the very mature applications and the very new applications," Mitchell said.
The mature, non-mission critical applications, including off-the-shelf products, are great first entries into Cloud for a number of reasons. Because your people know them, inside and out, they can focus on gaining a better understanding of how these apps "work" in a Cloud environment. Bulk email, for example, is often used for proof-of-concept pilots.
"The least mature applications are also good options because these typically have more of the problems that Cloud can fix," Mitchell said. "Maybe they have code that's a memory hog – Cloud can hide that. As long as they're not essential to your business operating model, these applications are great candidates for Cloud."
According to Reed, testing and development are also good options for proof-of-concept.
"Think about it – in an enterprise with 2,000 servers, 30 percent to 50 percent of those could be test servers, used for development, testing and staging," Reed said. "Moving this environment to Cloud can have a significant impact on cost with minimal risk."
Prepare Your Due Diligence
But, what about the fear factor often associated with moving to Cloud? The best defense is strong due diligence – asking for what you want and understanding how your potential providers operate.
"I think it's important to note that, even though it's Cloud, the work is still run out of a physical data center. Most providers position those data centers in various parts of the world. So, if you want to pick up server capacity closer to your customer, ask for it," Mitchell said. "For us in the travel industry, location is important, because it means faster response time for our customers."
If you plan on a significant capacity spike, make sure your providers can deliver.
"If you are a large player, you need to understand how the vendor is going to manage to that capacity. If you're going to scale from three servers to 2,000, you want to ask if you can do that with point-and-click provisioning or do you have to call them weeks in advance," Mitchell said.
But, what about combating those all-important security concerns?
"Ask how the provider runs its operations, how it monitors its system, how the technology in the background works," Mitchell said. "Get as much information as possible, then cover your risk in the contract – just like you would in a traditional ITO agreement with specified SLAs and penalties."
You can also consider adding software encryption, or running more sensitive applications in a private Cloud. The key is to determine how much risk you're willing to take, and then match that risk to the appropriate model.
Above all, start slow.
Although Cloud is a delivery mechanism, it has a lot of moving parts. So, the best advice is to ease into it. Take the time for your internal staff to adjust to the way this solution is deployed, monitored and managed. Then, continue up the application food chain as your confidence increases and business needs dictate.
"You have to work for a company that is willing to try new things, but you also have to be smart about your approach," Mitchell said. "It's like any other type of change. The more work you put into understanding the environment, the better your outcome will be."