Data Integration Blog -

Those moving to cloud computing understand a few basic things. First, they are hosting processes and data on servers they don’t control. Second, the use of cloud computing is driving the need to change the existing enterprise architecture. Finally, integration needs to be systemic to the enterprise architecture in order for that enterprise architecture to deliver reliable scalability and economic advantages.

So, what’s driving what? The interest in cloud computing is driven partly by hype, or the manage-by-magazine crowd looking to jump in feet first into a new and exciting trend, while the other half of the crowd have done careful study of the emerging cloud computing marketplace and have found a cloud computing provider worthy of their data and processes.

The movement and direction are clear. IDC predicts a compound growth rate of 20% per annum for cloud computing, set against the overall software market which is only growing at around 6% per annum. Moreover, a Gartner report shows global revenue from cloud computing will top $70 billion by 2013*. This includes the shift from on-premise to cloud-based providers, and thus the need for planning and architecture to make this happen.

The trouble with this shift is that most enterprises don’t have an architecture that’s ready for cloud computing. Why? Most clouds, including IaaS (infrastructure-as-a-service) providers such as Amazon Web Services, are sets of services or APIs and therefore need to communicate with services or APIs within the on-premise systems. Most systems are not service-oriented, and thus don’t expose data and behavior as services to interact with other systems, such as cloud computing providers. The end result is a diminished value of cloud computing, considering the inability for existing systems to communicate easily with other on-premise or cloud-delivered systems.

So, while you can do cloud computing without SOA, you really can’t do it well. Most of the thought leaders in the cloud computing space have already linked SOA to cloud computing. As more enterprises move to the “Cloud,” the use of SOA as an approach to alter the existing enterprise to take advantage of all systems, on-premise or cloud-delivered, will accelerate this year, and on into 2011 and 2012. This includes cloud-based data services, such as DataCloud2.

Core to this SOA evolution is integration, typically data integration. We saw early on as we began to leverage SaaS systems, such as Salesforce.com, that we somehow had to sync our SaaS-stored data back into the enterprise. With the rise of IaaS and PaaS (platform-as-a-service) clouds, critical integration systems are no longer just nice-to-haves, they are an outright requirement.

The reality is that infrastructure and development platforms must be well integrated back into the core enterprise systems to have the value to even justify their use. Indeed, most of those who deploy cloud computing providers typically do so without employing a business case. They will eventually discover that the use of cloud computing systems without core integration services deployed along with a SOA are actually counterproductive for the business because they failed to consider synchronization with core enterprise business processes and the risks around outages and outright failure of cloud computing providers. This is a dirty little secret about the use of cloud computing that most providers won’t tell you.

So, how do you approach cloud computing using SOA and the proper integration technologies? You must first assess your existing systems as to the existing issues, including interfaces and existing integration infrastructure…what’s working and what’s not. From there you look at the business case for driving SOA into your existing IT assets, including the use of integration and the potential value of cloud computing.

After you’ve created a business case and have an approved budget, it’s time to move into the planning stages, including a complete functional assessment of the data, services, and core business processes. From there you move to the “as is” architecture, looking at cloud computing as an architecture option that will typically only have value within the context of a SOA, leveraging core integration technology that can span intra-enterprise and between on-premise systems and cloud-hosted assets, including IaaS, SaaS, and PaaS.

The measure of success is a seamless integration between all major systems, cloud-based or on-premise, so much so that it’s completely transparent as to where those systems reside. Moreover, a successful system will have the ability to quickly change processes and core applications to accommodate a changing business, including mixing and matching the locations for data and processes on systems that provide the best value to the business, on-premise or “Cloud.”

There is no magic bullet here. SOA, integration, and cloud computing takes some time to get right. That said, the integration technology we have access to today can provide most of what you’ll need out-of-the-box, including the foundation of your SOA. However, keep in mind that SOA is something you do, not something you buy.

You can’t make the cloud work without SOA and good integration technology. Trust me on that one.

*This number does not include cloud-based advertising. From Gartner report: “Forecast: Sizing the Cloud; Understanding the Opportunities in Cloud Services,” Ben Pring, Robert H. Brown, Andrew Frank, Simon Hayward, Lydia Leong, March 18, 2009