This isn’t all or nothing – focus on the right cloud model for each application and not the entire data center

There’s a lot of discussion about why you should choose one cloud computing model over another and all of them miss the point entirely. This isn’t a mutually exclusive deal; it doesn’t have to be just one model chosen. In fact it shouldn’t be.

Data centers aren’t comprised of single types of applications. There’s custom applications, deployed sometimes on well-known packaged milton platforms and in other cases on open source or lesser known platforms. There’s packaged applications deployed on a variety of systems, and there’s applications that are running on aging technology that no one understands let alone is still supported except by one guy in the basement who tends to talk a lot like Milton from Office Space.

Hint: leave those last ones right where they are. Trust me, you don’t want to mess with it.

For the rest of your applications it may be the case that some are better suited to an SaaS environment than a PaaS, and in some cases IaaS is going to be your best option. It isn’t all or nothing, but rather it’s about choosing the environment that’s right for the application. Doing otherwise will simply cause you no end of grief and, in the end, might be as colossal a failure as the attempts to move applications not well-suited to SOA to SOA.

Square peg, round hole, you get the general idea.

So how do you choose which model is right for each application?


SaaS (Software as a Service)

Applications for which functionality, features, and general process have been well-established in the industry and require very little customization are ideally suited to SaaS environments. Sales force automation (SFA), customer relationship management (CRM), payroll functions, benefits management, and the like are all well suited to being delivered from an SaaS environment. In fact, many of these functions already are running in an SaaS environment as businesses have long understood the value of outsourcing the maintenance of such commoditized systems to someone else as a means to reduce costs and free up valuable IT resources to focus on providing business value rather than on reinventing the wheel.

And therein lies the big question for SaaS: If you’re considering reinventing a wheel and aren’t going substantially diverge from its function and features, seriously consider going the SaaS route.


PaaS (Platform as a Service)

PaaS sounds relatively new but really it’s an extension of hosted platforms that’s been adapted to fit the cloud computing model. PaaS providers focus on supporting a core operating system and set of application infrastructure platforms such as .NET or BizTalk or IBM WebSphere. Relative newcomers Google and Salesforce, too, offer options for building applications atop their proprietary platforms. You deploy applications that would normally be deployed in your own data center on those platforms in a cloud / hosted environment instead. The benefit is a reduction in licensing and management of those platforms, which again frees up IT staff to perform other tasks that offer more value to the organization.

If you’re comfortable with a particular vendor – because you’re locking yourself in with some offerings - and the strength of the provider and would prefer not to continue spending time and money on maintenance and systems’ administration and tweaking of the platform, PaaS is a good option for these applications. Both custom and packaged applications may fit into this category as many popular packaged applications are deployed atop popular application infrastructure platforms.


IaaS (Infrastructure as a Service)

When there’s no possibility that a cloud provider will support your particular platform, e.g. it’s running on OS/2 or an obscure web application platform, then you have only two options: rewrite the application to run on a platform that is supported or, more likely, choose an IaaS provider instead.

Applications that are highly specialized and have peculiar requirements will force you to examine the IaaS option because part of the profit margin equation for PaaS and SaaS is commonality and commoditization. Anything that steps outside those bounds is going to cost extra, and most PaaS and SaaS providers are wise enough to stay away from that kind of customization. IaaS offers almost limitless options for the organization, as it’s completely up to the organization to package up and deploy the applications . That offers the control, too, that may be necessary if you’re in need of special tweaks or third-party add-ons or plug-ins to the application infrastructure platform.


IT’S ABOUT THE APPLICATIONS
The decision regarding a cloud computing model really should be made on a per application basis because of the variables involved. Certainly if you’re looking to move toward a particular model as an organizational standard then you should do as much as possible using that model but realistically it’s likely not feasible for you to move every application to just one model. Also keep in mind that not all applications are suited to a public cloud deployment. Integration, data replication, compliance, and sensitivity to regulations and privacy must all be considered before moving willy-nilly to a public cloud environment.

You might be thinking that you could actually get away with deploying everything to an IaaS provider and therefore you could prove me wrong. But I’m betting that your organization already takes advantage of some form of SaaS – most do – and therefore you’ll never succeed in being a one aaS shop. But go ahead, prove me wrong, I don’t mind a little crow now and again (as long as you provide a tasty, gluten-free béarnaise to go along with it).

 

Follow me on Twitter View Lori's profile on SlideShare AddThis Feed Button Bookmark and Share