000

A few years ago, a consumer products company I’m familiar with committed to outsourcing its infrastructure and as much of its application portfolio as possible. Its strategic application decisions were heavily influenced by outsourced and SaaS offering, placing its CRM and several customer community sites in the cloud. With a stable set of hosted, cloud and SaaS apps, they were shocked to find that they didn’t talk to one another to provide a single view of their customers. Additionally, none of the vendors had an answer for the integration. This failing of the current state of cloud computing speaks to both adoption and hidden costs of fully embracing the cloud model.  This and other similar cases prove to me that integration is the primary barrier to systematic adoption of cloud computing.

This reminds me of the “best of breed” systems architecture problem in which an organization chooses to select the best application software for each of its business functions but is also saddled with integrating each of them together (often a tedious, custom project in itself). The need for this software integration created the enterprise application integration (EAI) middleware category.

One of my telecommunications clients took this approach which seemed like a good one when mitigated by a solid EAI vendor with adapters for each of the selected best of breed components - call center/CRM, order management, inventory management and financials. Despite some hard-core coding and customization of the adapters and some messaging and queuing debugging, the architecture worked well.

I think that the idea of connecting individual applications to a messaging or integration backbone could be applied to the cloud computing integration problem.  Let’s call it Cloud Integration Architecture (CIA).

Cloud Integration Architecture

ADP’s CIO Mike Capone and Chief Strategy Officer Jan Siegmund describe the importance of integration in both public and private clouds in their in-depth interview for PwC’s Technology Forecast on cloud computing.  For organizations who want to leverage as much in the cloud as possible, do they really want to perform the integration internally? While that is the reality today, there’s no reason that the integration function can’t be cloud based too.

The key integration points that a cloud integration architecture service would have to address are:

• User Interface Integration - aggregates and/or abstract cloud app provider UI APIs to help provide a consistent, customizable and modular UI,
• Data Integration - allows an enterprise to merge, purge and augment similar datasets from multiple cloud app vendors,and
• Function Integration - the most complex integration, involves gluing cloud APIs together into some kind of workflow or orchestration-like service.

What do you think?  Would you consider using a cloud application integration service? Or, would you build your own?

000