Service-Oriented Architecture Rationalization and Cloud Computing

I am going to present a talk on Service-Oriented Architecture Rationalization (SOAR) framework in the SOA Symposium (http://soasymposium.com/conference_agenda3.php) on Oct. 22, 2009. I'll also serve in a panel of Cloud Symposium - "Will the Open Cloud Become a Reality?" (http://www.soasymposium.com/panels3.php). Please feel free to contact me if you plan to attend, and I can provide you with a discounted registration rate or even a free pass.

Another interesting event this month is OOPSLA 2009 (http://www.oopsla.org/oopsla2009/program/tutorials#monday). I am scheduled to present a tutorial of Service-Oriented Cloud Computing Solution Development on Oct. 26, 2009. Please join the event and let me know if you have any questions/comments.

Smart Cloud

As we are marching in the journey of strategizing and implementing cloud computing, it is imperative to do it right as nobody can afford to repeat the mistakes and misfortune in the SOA growth. We got to be smart to systematically deal with cloud computing to avoid the hype and confusion. Some critical aspects are:

1. Set the right expectation - cloud computing is not a silver bullet as SOA was NOT as well
2. Use objective justification - not everything is a cloud service and not everything should be a cloud service
3. Take a hybrid approach - combine the top-down and bottom-up methods
4. Think beyond a technical perspective - technology matters, but business matters more
5. Aim at a good balance between strategic goals and tactical needs - stratactic is more practical

With collective intelligence and wisdom, we can build a smart cloud in a more cost-effective and faster way.

Cloud Computing Usage Patterns

As the Cloud Computing has been gaining more momentum, we need to be clear about where it is applicable for and where not, so that we can make sound decision to justify the adoption in the real-world situations.

I strongly advocate that we start with a usage model, where the business case can be established. A usage model describes the capabilities and functions within a stated context. It provides perspectives based on roles. A typical set consists of consumer, provider, partner, and mediator, as an example.

The usage scenarios can be grouped into usage patterns. For instance, the categories can be: basic business productivity solutions, contact management, task management, project management, event management, file management, financial management, service management, collaboration, social communities, CRM, HR, etc.

The Cloud Computing Use Case Discussion Group has been developing a whitepaper on use cases, which I contributed to in its infancy stage. Check it out. IMHO, what is lacking in the current version is that its taxonomy is unorthogonal and incomprehensive. The classification of use cases is not complete either.

Towards Unified Open Source Cloud Stack

As Cloud Computing is maturing, a variety of cloud solutions have been emerging. Open source solutions tend to provide more transparency and freedom to the users. There are several interesting open source cloud initiatives available.

Eucalyptus offers an open source clustering management for Xen images. Nebula provides an open source cloud computing platform, backed by NASA, which also uses Eucalyptus as the VM controllers. A number of cloud testbeds have been available for a while, such as Open Cloud Consortium, Open Cirrus, TerraGrid, PlanetLab, and EmuLab.

It is imperative to build a unified open source cloud stack with an open structure and replaceable components. It is critical that the open source stack should be standards-based, to consolidate and unify the proprietary stacks from vendors like Google, Amazon, and Microsoft. More importantly, this stack will provide a baseline for interoperability and drive standardization in the cloud computing space. Apache has been working towards this direction, and it can be foreseen that more efforts will be put in this area in the near future from the industry.

What is not Cloud Computing?

It seems that there is a trend that everything is becoming cloud-related, similar to the mania I observed in the early era of SOA at the beginning of this decade when literally almost every IT product marketed is labeled as a solution linked to SOA.

There have been so many definitions of Cloud Computing, but yet a consensus is still being built. More importantly, we have to be clear what is NOT Cloud Computing, so we won't fall into the trap of overusing/underusing/misusing the technology.

This comes down to the fundamentals - what is the core concept of Cloud Computing? What are the underlying principles? How to characterizes it? What are the key attributes and aspects? And what elementary constructs and models are the constituents of Cloud Computing?

I have been working on a publication addressing these questions. Please share your thoughts, views, insights, and opinions.

Service-Oriented Model-Driven Architecture Design for Cloud Solutions

I will present a tutorial on the service-oriented model-driven architecture design for cloud solutions in the upcoming International Conference on Web Services (ICWS 2009). Please join the session to explore the state-of-the-art approach to effectively developing cloud services in a systematic fashion.

Cloud Engineering

As the emerging cloud computing is getting more attention nowadays and is expected to mature over the next few years, a question is raised as to whether the engineering of cloud computing is a disruptive innovation. In other words, is the discipline of cloud engineering a convergence of the traditional IT-related engineering practices, such as software engineering, systems engineering, web engineering, service engineering, and platform engineering? Or is the cloud engineering in its own domain, dealing with the unique challenges faced by the cloud computing, like multitenancy as an example?

Your views, insights and opinions are welcome. FYI, I chair a special panel in the upcoming International Conference on Web Services 2009 with experts lined up to discuss this specific topic in the session. Come and join us in the discussion.

Cloud Computing Body of Knowledge (CCBOK)

As the Cloud Computing is maturing, it becomes important to establish Cloud Computing Body of Knowledge (CCBOK), as a written guide to the collection of cloud engineering lifecycle knowledge reflecting best practice and lessons learned, providing a framework that describes the areas of knowledge, with associated competence, activities, tasks and skills required. Here is a jump-start with a cloud metamodel:



Comments are welcome. More to come...

Contact Tony Shan (tonycshan@gmail.com) for more info.

Cloud Taxonomy and Ontology

There have been a lot of discussions on cloud computing. The fundamental question is how the cloud services are logically organized from a capability perspective?

A structured way is needed to provide a taxonomy to classify a vareity of XaaS offerings. Here is a dome model for cloud services. Detailed definitions with examples are being constructed. Stay tuned...


Contact Tony Shan (tonycshan@gmail.com) for more info.

Cloud Environment

The overall Cloud Environment is illustrated in this diagram. It consists of 4 integral parts:

1. Foundation
2. Service Integration & Management Platform & Lifecycle Engineering (SIMPLE)
3. Cloud Service Factory
4. Composite Clouds

Contact Tony Shan (tonycshan@gmail.com) for more details.