Cloud Computing Adoption is Accelerating
Who wouldn't be interested in extensible functionality and computing resources at an attractive, pay-as-you-go price?

The economics of Software-as-a-Service (SaaS), Platform-as-a-Service (PaaS) and Infrastructure-as-a-Service (IaaS) are just too compelling to pass up.

However, because today's standalone cloud application may prove to be tomorrow's integration nightmare, banking on cloud computing is not a recipe for a good night's sleep.

The Coming Cloud and On-Premise Integration Nightmare
Have you ever awoken startled from a bad dream where you were taking a final for a class that you had never attended?  Not being prepared for a major challenge is a terrible feeling.

Enterprises implementing cloud solutions today are just about to wake up--startled by the fact that their siloed cloud solutions now need to be integrated with their on-premise systems and they are not prepared for this challenge.

Cloud and on-premise integration can take two forms.  From a business process point of view, the integration needs to enable end-to-end business processes that integrate functionality from transaction systems inside and outside the enterprise.  The lead to revenue business process is an example where on-premise Unica marketing automation activities needs to flow to the cloud-based Salesforce.com sales automation and then back inside to the SAP order management systems.  To do this, transactional integration is required.

Business analysis integration is the other form of cloud and on-premise data integration on the near horizon.  Across the same Unica, Salesforce.com and SAP on-premise and cloud systems, a wide range of useful marketing campaign, sales effectiveness and customer profitability analyses are required.  To do this, data integration is a must have.

The Trend Is Not Your Friend
In the simple example above of an implementation of one cloud application, Salesforce.com, resulted in five cloud and on-premise integrations.

• Unica to Salesforce.com transaction integration to promote developed leads into the sales pipeline
• Salesforce.com to SAP transaction integration to progress sales activities into revenue transactions
• Unica, Salesforce.com and SAP data integration for marketing campaign analysis
• Unica, Salesforce.com and SAP data integration for sales effectiveness analysis
• Unica, Salesforce.com and SAP data integration for customer profitability analysis

Fast forward a few years when your enterprise has added dozens of cloud solutions to the hundreds of applications in your existing on-premise portfolio. How many point transactional and data integrations will be required then? Dozens? Hundreds? Thousands?

Cloud to On-Premise Data Integration Is Difficult
Not only are the volumes daunting, these integrations can be challenging to build and maintain.  Extending your old integration methods won't work.  Take data integration for instance.  There are multiple new requirements that must be addressed.  Traditional direct database queries and ETL script techniques don't work outside firewalls.  So you will need to adopt new data integration approaches such as data virtualization that are architected for more loosely-coupled integration scenarios.

In addition, when you integrate data from a SaaS provider such as Salesforce.com, you will need deep knowledge of Salesforce's API and how to do on-demand, rather than batch-mode data integration. This will require new cloud solution expertise as well as specialized integration tooling.

Further, because you now need to query data through a firewall across the Web, the security challenges expand.  New authentication, authorization and encryption techniques must be adopted.

Can This Nightmare Be Avoided?
By now you can see that massive numbers of integrations and a significant change in integration methods and tooling are inevitable as cloud computing adoption moves ahead.  The question is what you should do about it.

To break away from the point-to-point integration mentality that has dominated IT architectures for many years, new thinking is required. In June 2011, Gartner wrote an interesting report that provides good guidance on this issue. Data Integration Hubs: Drivers, Benefits and Challenges of an Increasingly Popular Implementation Approach considers data and transactional integration in a more holistic way that leverages canonical approaches.  The hub approach instead of point-to-point provides consistency and reuse across sources and consumers.  Forrester, with their Data Virtualization research, provides similar counsel.

Beyond new thinking, new methods and tooling are also required.  Enterprises need to evaluate and adopt new integration solutions better tuned for cloud and on-premise integration requirements.

Data Virtualization Users Can Sleep Well
On the data integration front, data virtualization users are fortunate because they can easily extend their on-premise data virtualization platforms to flexibly integrate on-premise and cloud data.  Some of the advantages of using data virtualization for cloud and on-premise data integration include:

• Data virtualization's powerful data abstraction tools simplify complex cloud and on-premise data, transforming it from native structures and syntax into easy-to-understand, reusable views and data services with common semantics. This helps avoid point-to-point integrations.
• Data virtualization's loosely-coupled data virtualization layer, rapid development tools, automated impact analysis and extensible architecture provide the agility required to keep pace as new cloud solutions are brought on board.
• Industry-standard APIs for wide range of cloud based sources and consumers simplify and speed new development. Advanced data virtualization platform providers include vendor-certified adapters to integrate popular SaaS applications such as Salesforce.com and on-premise applications such as SAP and Oracle E-Business to further simplify more popular integrations.
• Data virtualization services operate both on-premise and across the Internet without any extra work. Further existing data virtualization security models ensure proper authentication, authorization and encryption both on-premise and in the cloud. This removes deployment risk.
• Because data virtualization accesses, federates, abstracts and delivers queried data on demand need to purchase additional on-premise or cloud-based storage for ETL and data consolidation.

Think Ahead So You Can Sleep Well Too
The economics of the cloud are compelling.  But without proper foresight, today's standalone cloud application may prove to be tomorrow's integration nightmare.

Don't let it happen to you.  Instead evaluate new approaches now, including data virtualization which is especially well suited for cloud and on-premise data integration.