David S. Linthicum | Why SOA matters in migrating to the cloud
Connecting state and local government leaders
Service-oriented architecture provides an ideal platform for building a private cloud, which first requires breaking down your systems into services, processes and data components.
On April 29, I spoke at the Cloud Computing Summit in Washington, on a panel entitled "Extending IT to the Cloud" along with the new chief technology officer of the federal cloud, Patrick Stingley. After the panel, I did a presentation that looked at how you approach cloud computing using service-oriented architecture.
What I found most interesting at the conference was that cloud computing is a very hot topic in the U.S. government. With the new cloud-friendly administration, those in government information technology feel compelled to look to the clouds for their next-generation IT needs. However, although they talk about public clouds, they are actually defining and building private clouds, which are SOAs with the addition of virtualization and self-provisioning.
So why is there buzz in the government around private clouds? Considering the needs of the government, most agency IT managers find they are not yet ready for public clouds. Control, security and interoperability are key issues. IT managers love the model of public clouds but just can’t make it work within their architectural requirements.
Those requirements provide the government with an opportunity to build the mother of all private clouds. Take a moment to contemplate a private government cloud that is shareable among agencies, within a protected network environment. Now consider the amount of efficiencies you could drive into government IT with shareable and on-demand core computing resources, and the case for a huge private cloud for the government becomes compelling.
You can think of private clouds as a collection of systems that provide sharable storage, database and processing services — much like public cloud computing providers such as Amazon and Force.com do — and reside within firewalls and data centers. They are a group of systems in which the processes are managed through a virtualization manager that treats many distributed computers as if they were a single virtual server. Another core difference is that you typically build private clouds by defining a SOA with self-provisioning and virtualization, whereas in the world of public clouds, you can build on infrastructure that already exists.
The advantage of taking advantage of this kind of architecture is that we’re good at utilizing existing computing resources because all processes are distributed automatically among any available resource. Utilization of a traditional computing model server, as a percentage of its capacity, is typically in the single digits because the applications are bound to particular physical computing resources, such as a single server. However, when that server is part of a virtualized environment in which the application runs on a virtual server that spans many physical computing resources, the utilization of the servers goes way up, perhaps 20 percent to 30 percent on average. Thus, you can provide more processing power within a shared virtualized environment, using fewer physical computing resources. That also means the costs of hardware, software and administration should go down.
SOA comes into play because those who use private clouds, or virtualization, typically break down new and old applications as services, processes and data and address each as an architectural component that may be freely distributed in the private clouds. When you consider the value of private clouds, or public clouds for that matter, SOA is the approach that makes the most sense. It’s the approach that many in government are deploying as they move toward private cloud computing.
For the government to make a successful move to private cloud computing, IT leaders need to break existing systems down to their services, processes and data components before the move. Therein lay the opportunities to leverage a platform that is cost-efficient and fix aspects of the architecture that can drive processing efficiencies. SOA best practices and approaches are the best path for government IT to accomplish these goals.
NEXT STORY: Ready for a closeup: GSA debuts on YouTube