Staging or Direct – SaaS Integration Best Practices?

I wrote a post today on the Informatica Perspectives Blog about the differences between point-to-point business application integration and staging enterprise data. It’s a topic that many organizations are discussing as it relates to cloud data integration. In order to avoid data fragmentation and data inaccuracies that ultimately result in decreased SaaS application adoption, a lower TCO, and ultimately poor business decisions, data from these cloud/SaaS applications must be integrated with each other, as well as with on-premise systems.

But for most companies, “the cloud” has  fundamentally changed the process of evaluating, purchasing, deploying, and managing software. Line of business (LOB) is commonly driving the technology purchase and implementation process, and in some cases, the central IT organization is being bypassed altogether. In larger organizations, LOB is attracted to faster time-to-results, subscription pricing, etc. and they’re adopting cloud applications to avoid the delays often associated with centralized corporate IT involvement. As this CIO.com cloud trends for 2010 article noted:

“Politics will drive decisions:  Cloud-based services have demystified compute services and increasingly business executives will ask IT managers about the cloud.”

With this in mind, are there new best practices emerging around cloud data integration? If so, what role does IT play? And if you’re in IT, do you provide the business with self-service access to production back-office systems or do you stage the data?

I’m interested in hearing about how this works in your company. Got any best-practices to share? Here’s a whitepaper on the topic that might be useful (registration required).