I’ve come to the conclusion that it’s all of the above.

 SOA is an architecture paradigm and a set of principles to guide service-based applications, platforms, and even enterprises (SOE).  Cloud Computing is a computing platform paradigm that consolidates physical services into virtual shared services (infrastructure).  They don’t conflict or compete.  And they don’t enable each other.

SOA and Clouds Complement Each Other

They can, however, complement each other to create a whole that’s greater than the parts.

The confusion continues, however, and has grown to be a major topic of debate.  There are on-going discussions about the relationship between Cloud and SOA – claims that one requires the other; suppositions that SOA is dead because Cloud is the new service platform; analysis of the criteria that would determine if a business should adopt cloud or SOA.

For me, it became a serious issue recently when I was tasked by a client to address the question. I was asked to clarify the two paradigms’ value propositions, create a vision for how they can be leveraged together to create a service platform, and develop a roadmap to get there.

As an outgrowth of that initiative, I’ve started a series of posts on StrategicITArchitecture.com. My thesis is that Cloud and SOA can be combined to create the Ultimate Enterprise Service Platform. The series will frame the issues and present a set of recommendations and framework for developing a roadmap.

Check it out, if you’re interested, and please comment if you disagree, agree, or have suggestions.  You can also take a look at a scaled-down version of a Cloud/SOA presentation that I prepared for a client.

I’ve just posted some guidance for developing a Cloud Computing Roadmap on my Strategic IT Architecture Web site. Obviously, it’s a complex topic and there are many, many variations on the routes you might map out.

One approach that I didn’t mention in that post – space is limited in a blog posting – was an approach that mitigates and contains risk, while still delivering a solid foundation for internal Clouds and near-term ROI and operational benefits.  Server platform virtualization.

Virtualization is a mature and proven platform architecture that greatly improves asset utilization and can significantly enhance operational efficiency. Clouds are based on a virtualized platform.

In order to fully evolve into a Cloud, and provide all of the benefits that Clouds offer, you will eventually need to enhance the virtual environment with some additional capabilities – e.g., self provisioning, dynamic resource allocation, measured/metered usage. You’ll also want to make some enhancements to the platform architecture to better accommodate multi-tenancy, elasticity and other powerful Cloud capabilities.

But as a solid first step toward a Cloud Computing target, and as a means of containing risk – including the risk of moving core applications to an external, still unproven commercial Cloud – consider implementing a virtualized computing IT ecosystem.

One additional technical point: As you develop your implementation plan for a virtualized platform, you do need to consider what kind of “Hypervisor” you’ll use. The most critical decision is whether to use a Type 1 (bare metal) product or a Type 2 (hosted) Hypervisor.  Type 1 provides maximum performance because it executes on the hardware (bare metal) and provides its own operating system support. Type 2 is installed as a guest operating system on a hosted system, so it sacrifices some performance because it’s running in a layer above the native OS.

Here’s a quote from a briefing that one of my clients used to kick off their Cloud strategy discussions. The definition is valid, but only for one specific model of Cloud Computing – public (commercial) clouds.

“Cloud computing refers to the practice of purchasing computer services that are stored and maintained by a third-party contractor, instead of housing all the equipment and software on-site.”

When this slide appeared on the screen, 10 minutes into the kickoff session, most of the leadership team reacted immediately. Their response, based on the misperception created by that focused and incomplete definition, was… we do not, and will not outsource our IT platforms to anyone!

It took a quickly improvised white-board session to clarify the opportunity and to get this initial strategy session back on track. There are tremendous benefits available from the use of an appropriate Cloud Computing model and its enabling technologies.

Examples of the strategically significant Cloud models are:

• Public Clouds (defined above) – examples include Google (e.g., GAE), Amazon (e.g., EC2, S3), and Microsoft Azure.
• Private Clouds – managed for the use and benefit of a single organization. Significant benefits from consolidation and virtualization of platforms and operations.
• Community Clouds – shared by multiple communities of interest. Benefits from pooling resources and operational costs.
• Hybrid Clouds – Consisting of Private Cloud(s) and augmented by integration with Public Clouds for “cloud bursting”, auxiliary data storage, or for access to commercial applications.

One common example where lack of specificity in the definition of the Cloud results in a misleading conclusion, the recurring argument that security is a major inhibitor to adopting Cloud Computing. This concern is only valid (to some degree) when referring to commercial Cloud vendors who have not, yet, reached security compliance. Growing maturity in this space and lessons learned have shown that multiple data centers, consolidated into a Private Cloud can be hardened, secured and managed to a much higher level and at lower cost than is possible across multiple data centers.

Bottom line: the potential benefits are significant. Just follow well-proven best practices – understand the problem space, clearly define goals, identify the risks, assess benefits (ROI being only one measure), identify the best, most appropriate solution (e.g., what’s the best Cloud model) – and you can craft an effective roadmap to reach your target. Significant benefits await!

If you’re interested, I’m discussing a more detailed approach to developing a Cloud Computing Roadmap on my Strategic Architecture blog at http://StrategicITArchitecture.com.

There is, of course, a great deal of value to be gleaned from the cloud movement – both from public cloud services and by implementing private/internal clouds. As cloud vendors and services mature, they will be a valuable addition to the IT landscape and will fill a variety of roles (some as core services and others as utility services). But enterprise users should not rush to try to realize the promised benefits without a clear strategy, realistic planning and – above all — risk management.

Just a brief comment about the etymology of the term “cloud. We used to use “cloud” as a symbol for some entity out on the network. We weren’t concerned with what was in that cloud (generally, it was a black box). Someone else was providing results/output from that we needed. It didn’t matter how the functionality was implemented as long as it delivered what we needed at the edge (contracts and other means of assuring the usability, quality and usefulness had to be in place, of course).

The cloud in “Cloud Computing” as we use the term today is quite different. These Clouds need to be grey boxes, not black boxes. We need to understand much more about their construction and how they operate. I strongly recommend that if you’re planning to host core processes or data in/on a cloud that’s managed and controlled by another party… you need to exercise due diligence and establish clear, contractual agreements on SLAs, ensure that security is appropriate for your use, and understand your data and process migration options — and that’s just for starters.

I’m elaborating more on this and other related topics to create a clear framework for a Cloud Computing Roadmap in my blog at http://www.StrategicITArchitecture.com.