Wednesday, August 20, 2008

Semantics & SOA - part 1 "Survey"

Over the next month we are going to take an in depth view of how Semantic Integration is being applied to SOA. The notion that the Semantic Web and SOA might be linked began to emerge in 2005 & 2006. Since then it's been gaining some momentum. Keep in mind though that "Semantic Integration" & "Semantic IT" do not by their nature narrowly confine themselves to SOA or the Web. However, Services Oriented Architecture is a natural, ready-made proving ground for what we're describing on this Blog.

This first post reviews some interesting resources on the web that illustrate the genesis and evolution of the combination of Semantics and SOA.

Conceptual Framework:
This is an excellent briefing which lays out basic concepts and potential applications for SOA, presented by the Chief Scientist at Top Braid, a semantic web technology provider.

A concise overview from the Zapthink crew.

The wikipedia take, or one of them anyway...

Semantic coupling

An excellent assessment of why many SOA efforts are falling short...
excerpt "Unfortunately, syntactic coupling is the easy part. Semantic coupling is the harder part of the problem, and SOA does little or nothing to address this challenging issue. Semantic coupling refers to the behavioral dependencies between components or services. There’s actual meaning to the interaction between a consumer and a service."
We will be taking a much closer look at this issue soon...

Interesting Applications:

More links Courtesy of Mills Davis:
In our next post we will try to outline the architectural imperative for using "Semantic SOA."

copyright 2008, Semantech Inc.

Saturday, August 16, 2008

Semantics & Governance

Someone asked me recently whether or not there some type of tangible relationship between Semantic Integration and governance - it was an excellent question and the answer is a resounding "Yes."

Let's approach this at each of the levels where it applies:

Level 1 - Understanding the Nature of Governance at a Conceptual Level

Level 2 - Using Governance as an enterprise process medium

Level 3 - Using Governance as a foundation for detailed policy management (at the technical level)

While much of the current dialog in IT has focused on SOA Governance, IT or Enterprise Governance has a much wider scope than Services Oriented Architecture. And of course all of this extends into business areas or domains that go beyond IT. One of the great things about Semantic Integration is that at its heart it is not limited to Information Technology per se, Semantics provide the foundation for Every aspect of an organization, not just its IT capabilities.

So let's return to level 1 - our first problem is that most organizations don't currently agree on what Governance is. As evidenced by our discussion thusfar, Governance can be directed at many specific areas or be viewed holistically across them all. Semantic Integration helps us determine the core definitions and relationships in this conceptual competition - at the heart of all Governance definitions lies the need to conduct "Oversight." The corollary to that is the ability to engage in active remediation of issues discovered within the scope of oversight. These are both 'management' activities.

Our next post will explore level 1 Governance in more detail...

Copyright 2008, Semantech Inc.

Tuesday, August 12, 2008

The Here & Now of Semantic Integration

I've read some articles again this week explaining that the promise or potential of Semantic Integration is still a bit far off from being able to engender significant impact to the enterprise. Perhaps, it is worth examining the subject more as a practice approach rather than a standards or technology dependent facilitation medium.

By that, I'm simply reiterating that "Semantic Integration" is not synonymous with the "Semantic Web." There is a quite a bit that can be done right now with the techniques being described here (in concert with a limited but growing set of Semantic tools, although they aren't always required). As our previous post highlights, one of the first places where this can occur is with Enterprise Architecture. Anyone who has ever spent the time reading through DoDAF data models or building meta-models in Metis can testify quite convincingly that the entire field of EA is essentially one giant semantic exercise. I still tend to describe myself as an Enterprise Architect or Integrator; it was specifically because of my roles that I was drawn to Semantics as a way to solve a number of persistent and related challenges.

Take that further, what underlies UML, design patterns and J2EE language syntax ?... more semantics. It's already here, it's already pervasive within IT, but what we haven't done yet is fully appreciate this linguistic tower of babel we've created (and unlike the real world where the approximate number of languages is more or less stable with several universal translation modes, IT adds new semantics EVERY DAY).

Semantic Integration begins with this realization - that all enterprises represent ecosystems or cultures with unique semantic perspectives. Furthermore these virtual ecosystems are unlimited in regard to how many relationships may develop (we not only have individual entities within communities and communities within communities, we also have dynamic re-alignment both externally and internally). So step one is simple, define our enterprise perspective - this is the foundation for all other IT efforts and can be done using any number of tools starting with pencil and paper. How many data standardization or MDM efforts skipped this step and focused on defining core data elements that were already biased toward DBMS paradigm or another?

Semantic Integration is IT's first, future and favorite natural language activity. Eventually all discovery and all IT-related capabilities will be managed using natural language rather than XML, SQL , J2EE or some other technical proxy language or syntax. The sooner that occurs, the sooner that we will recognize the benefits both of IT and Semantic Integration. So how does one get started - it seems daunting, well maybe it is if you are too tied to one architectural perspective or language but if you can view this situation with an open mind there are several ways to get started.

(Next Post - Getting started with Semantic Integration...)

This view illustrates a real-world example of how Semantic Integration was introduced into a large enterprise without requiring a radical influx of Semantic Technology...

Copyright 2008, Semantech Inc.

Monday, August 11, 2008

Semantic Integration & Enterprise Architecture

In many ways, Enterprise Architecture (EA) is as misunderstood as Semantics. Although EA has been practiced across a much wider community of IT professionals for a longer period of time, it still suffers from an identity crisis. Is EA the mandatory precursor for model driven development, or is it part of a bigger picture and if so, what is that picture?

It is my contention that the reason Enterprise Architecture is still misunderstood in many quarters and often unsuccessful in practice is precisely because it does exist within the context of a larger picture. All too often, that larger picture is simply ignored leaving those executing EA projects somewhat perplexed as to find meaningful ways to make their efforts relevant to the organization sponsoring their efforts.

Enterprise Architecture represents a practices, tools and techniques which have evolved to help define the nature and state transitions of an organization from an IT perspective. The business view of the enterprise is of course included within this perspective but EA at its heart is and always has been driven by technology. The EA perspective is of a virtual entity, perhaps even the cyber-identity of an organization. This view allows those who manage and maintain complex system of systems ecosystems through exploitation of a holistic characterization of all capability elements. However, like most things in IT, EA as a discipline suffers from a lack of clarity regarding its core principles and approaches. In other words, there is no agreed upon definition for any aspect of Enterprise Architecture right now.

Seeing a big picture - an example of a "Meta-EA view"

So, Semantics and EA come together on many different levels; first in the need for one to clarify the other, secondly in the ability to build that larger context and big picture view of where EA fits along with all of the other aspects of IT. To understand the complexity of the problem, it is worthwhile to capture some of the competing definitions that one current finds associated with the term “Enterprise Architecture:”

  • Enterprise J2EE Architecture – This has been formalized within the curriculum and certification paradigm of the Sun Java Enterprise Architect designation. [ ]
  • Department of Defense Architecture Framework (DoDAF) – A wholly separate framework paradigm from FEAF.
  • ToGAF, Zachman Architecture Frameworks – Commercially driven approaches.
  • Unified Modeling Language – Often included within the context of other approaches but also often used standalone to manage EA efforts.
  • Agile Enterprise Modeling & Architecture – Much more application development focused, generally less formalized than strict UML.
  • Other Technical Specific Architecture – This is a long list and can include things such as .NET, Web 2.0, and literally 100’s of other software products.
  • Service Oriented Architecture – This often now is bundled with business modeling using BPEL.
  • Semantic Architecture – Yes, there a few people using this now although mostly within the context of semantic web technology or pure Ontological development.

So, what then is Enterprise Architecture, really? Is it a framework or set of frameworks, is it product specific skills, is it language specific skills, is purely technical or partially business focused, is it the metamodels and notation language used to characterize design?

Semantics allows us to integrate Systems & Architectures

The simple answer is that EA has been and will be what it needs to be to those who need it. There is no one approach now because no one approach will handle all of the related duties that architects are saddled with. The more important question has always been, will we find a way in which we can coordinate the various types of EA activities. This is an exact corollary to the questions that helped launch EA as a discipline (of set of disciplines) some twenty to thirty years ago. At that time pioneers in EA were looking for a way in which they could combine and coordinate the complexity of their systems environments. So, now its seems that EA has lead us to meta-integration. And what is the one approach we now have to tackle the problem at the top of complexity pyramid – Semantics.

copyright 2008, Semantech Inc.