As cloud computing begins to gain traction within the enterprise, there seems to be a refocus on data integration. Indeed, even with the down turn in the economy and thus budgets, many enterprises are moving towards cloud computing to save money, but are finding that data integration is still a core requirement. I'm finding that many did not figure that out until resources were being stuffed into clouds.
What's occurring is that data is being relocated into cloud services, either through database-as-a-services or applications-as-a-service. In both instances you need to create a strategy for syncing that data back into core enterprise systems.
The patterns of data integration within an enterprise work very much like data integration from cloud computing providers to the enterprise, with the complication of the enterprise having to leverage interfaces they don't control. Most cloud computing providers provide APIs for information consumption and production, however the cloud computing user is still charged with handling the transformation and routing of the information as it flows from the enterprise systems to the cloud-based systems. That typically means some integration technology.
This is not hard, we've seen this movie before. However, many enterprises are not accounting for the costs and complexities of doing data integration between the clouds and the enterprise, and since they are typically moving to cloud computing providers for the cost savings, you need to consider that as well. In some instances, moving to a cloud provider could actually cost more, when considering the data integration costs.
More evidence that data integration is systemic to everything.
Dave -
You are correct that moving the servers into the cloud exposes the real challenge, data integration. Data is a mess. One that gets worse by the day.
Data virtualization, Abstraction layers, SOA Data Services, Private Data Clouds, etc. are all variants of the loosely coupled, semantic data layer that enterprises need to insert between their hundreds of siloed sources and thousands of consuming solutions if they want any chance at agility, cost reduction, simplicity and more that cloud computing promises.
-Bob Eve
To David’s point that “we’ve seen this movie before”, I would agree, though in some cases the people involved in this particular migration from premise to cloud may not have been involved in previous integration projects, or may not even be in an IT organization. We’ve seen over the past few years that as more organizations adopt SaaS applications, such as Salesforce.com CRM, Taleo HRMS, or Concur Expense Management and other cloud based applications, the line of business administrators have become the owners of the applications, not the traditional IT people who may have “seen the movie before”. Further the engineers writing the new generation of SaaS and Cloud-based applications often write them to be completely self-contained, with no initial thought or architecture to integration with a client’s back office systems and databases. To address the needs of both of these groups who are not familiar with traditional integration tools, a true cloud-based integration solution is required that is designed from the ground up to be easy enough for a SaaS administrator to use and capable of connecting the data in these isolated SaaS applications with the databases and systems inside the corporate firewalls. You’ll find some interesting case studies of companies that have successfully used a new generation of On Demand tools to do just that at http://www.InformaticaOnDemand.com. The industry needs to provide purpose built solutions to this new generation of administrator and developer who work and live in ‘the cloud’.