From business basics like email to specialty uses like high-performance computing, public cloud services continue...
By submitting your email address, you agree to receive emails regarding relevant topic offers from TechTarget and its partners. You can withdraw your consent at any time. Contact TechTarget at 275 Grove Street, Newton, MA.
to beckon. But how can you be sure you’re not hearing a siren’s call and wind up doomed by a bad IT decision?
To ensure that your experience doesn’t parallel the Greek myth, you must start with applications that are highly repeatable, said James Staten, principal analyst at Forrester Research Inc. “What I mean by that,” he explained, “are applications that can run on a standardized, commoditized infrastructure or can be replaced by another fairly standardized application.”
This rule applies whether you’re considering the public cloud as a place to install your applications using Infrastructure as a Service (IaaS) or Platform as a Service (PaaS) options or as a way to provision Software as a Service (SaaS), Staten said.
Customer relationship management (CRM) makes a perfect cloud application starting point for enterprises, said Douglas Menefee, CIO at Schumacher Group, an emergency management firm in Lafayette, La. “It’s so mature and proven; anybody should find it an easy transition,” he said.
For small and medium-sized companies, email cloud services might be a better first step, Menefee added. Schumacher Group already runs about 90% of its applications and 85% of its processes inside the public cloud, he said.
The company started out with CRM from Salesforce.com and has since added in cloud services for email marketing, email and calendaring, learning management systems, human resources management and many other applications. Most recently, Schumacher signed up for a budgeting and forecasting service from Host Analytics, Menefee said.
“We brought up this cloud solution in less than six weeks -- amazing, considering that by the time we moved into the testing phase of the cloud-based solution, I wouldn’t have even had the server provisioned inside my data center, had I decided to go on-premises,” he said.
Despite the preponderance of cloud applications at Schumacher, Menefee said he doesn’t have a cloud mandate. “But what I've done is not rule cloud solutions out when I have a business need. And nine times out of 10, the cloud solution has beaten out the on-premises, based on factors such as time to market, cost and ease of integration,” he said.
“Every decision should boil down to costs and benefits,” said Steven Peltzman, CIO of New York's Museum of Modern Art (MoMA).
Security, lack of control and privacy issues are the obvious places when it comes to IT concerns about using the public cloud. But you have to weigh those against the benefits offered by cloud computing and then compare that complete cost/benefit picture to running the application in house, Peltzman added.
One way to ensure you’re starting your foray into public cloud computing in the right place is to lump each application into one of two buckets: opportunistic or systematic, said Eric Knipp, a principal research analyst with Gartner Inc.
Opportunistic applications are those that call for quick development cycles without too much concern for manageability or quality of service (QoS). These requirements aren’t big focuses because this type of application, such as a special marketing website, doesn’t have a long life expectancy, said Knipp.
As an example, Knipp cites Best Buy Co.’s use of the Heroku Platform-as-a-Service cloud offering to support IdeaX, an online community for customer and employee interaction. Embracing PaaS has allowed Best Buy to focus on application development while leaving operational concerns to Heroku, the company states on its website.
Systematic applications, with enterprise resource planning being the prime example, fall on the other end of the spectrum, Knipp said. These are the types of applications that require mainframe-like QoS and strong management because they’re mission critical and long term in nature.
“It’s tough making an argument for moving systematic applications into the cloud today, mostly because strong enough service-level agreements aren’t available yet,” Knipp said. “But for opportunistic applications, it’s game on as far as I’m concerned.”
Staten agreed, although he uses the terms “elasticity” and “transiency” rather than opportunistic to identify cloud-suitable applications.
“Can your applications scale up and consume more resources when busy and then scale back down when not? If they can do that, cloud economics favor you,” he said. “Likewise, if your application is only going to live for a few hours, you can’t beat the value of the cloud.”
Websites designed to scale with load balancing fall under the elasticity umbrella. So do high-performance computing (HPC) applications that are typically deployed using a grid model for tapping into loosely coupled compute resources.
“HPC applications can easily spread out across 20, 10,000 or 20,000 nodes, so they have great elasticity. But everybody’s been constrained by how big a grid they can build while in the cloud. There’s nothing to constrain you,” Staten said.
When you think of transiency, doing test and development in the cloud makes perfect sense. “You’re going through the cycle of working with an application that will eventually end. That fits the transiency world beautifully, as long as it’s a simple application that can go on standardized hardware,” Staten said.
Project-based collaboration applications -- those used by a product launch team or financial number crunchers during a quarter’s closing -- are good to run in the cloud, too, Staten added. These include wiki sites, SQL Server, a storage volume for file sharing and more. “There’s a variety of things you can do in the cloud under the collaborative heading,” he said.
When elasticity and transiency characterize an application, developers are adopting IaaS and PaaS cloud services at a fairly rapid pace, Staten said. “They're trying to get a job done and they see the cloud as a means of doing that very quickly.”
SaaS, with its “take it or leave it” nature, can be a harder sell within a large enterprise, Staten said. SaaS in the enterprise tends to depend on maturity of the offering and the commodity classification of the application category,” he said. Gartner mainly sees collaboration, CRM, email, human resources and security SaaS offerings as the bigger enterprise plays, he added.
As MoMA’s Peltzman says, “For us, the cloud [can be] an extremely fitting solution. Our resources are precious and need to be focused on our innovating to address our mission, not on running email or developing of customizing large applications.”
Beth Schultz is a longtime IT writer in Chicago. You can reach her at firstname.lastname@example.org.