Enterprise Architecture (EA), as defined by the Zachman Framework in the 1980s, has aimed to be a guiding light for large organizations seeking to align their business strategy with IT capabilities. However, as the world has evolved, so too have the challenges in this field—particularly around scale, integration, and decentralization.
The word “architect” originates from the Greek arkhitekton, combining arkhi– (chief) with tekton (builder or carpenter), meaning “master builder.” An architect is expected to be an accomplished builder, earning the title of “master” rather than simply being a “designer” or someone who drafts pretty diagrams.
In this sense, the EA role evolved into what many regarded as a “Master Planner”, an ominous title, however, this single persona, created a centralized, rigid structure that isolated architects from the rest of the business. This model, along with the legacy tools that supported it, struggled to deliver the insights and agility required to navigate modern, distributed architectures. Gregor Hohpe captures the diversity of architect personas in his entertaining book “The Software Architecture Elevator” 1, but it is typically referred to as the “Ivory Tower” model.
When I first encountered EA, I noticed field was shifting away from the old-school ideas of Zachman frameworks and solutions centered around Visio and SQL that promised more flexibility and insight from structured data. The advent of Agile development, global cloud-based infrastructure, and Software-as-a-Service (SaaS) business models, new EA tools were disrupting the market with both more modern approaches and higher integration capabilities.
As I delved deeper into the world of EA and these old and new tools, one persistent question remained: If we are to avoid the “Ivory Tower” centralized command-and-control approach that these tools continue to promote, what is the solution? Are there tools that can support a decentralized—or better yet, a federated—approach to Enterprise Architecture? Could such an approach, one that distributes architectural responsibility and fosters dynamic collaboration, be the key to supporting modern, large-scale, decentralized organizations?
In this post, I reflect on this pressing question, drawing from my personal experiences and observations of the industry players, my perceptions of the shortcomings of both legacy EA tools and emerging solutions, and consider whether the rediscovery of Ontologies and Semantic Web technologies like RDF and OWL could be the key to achieving a truly decentralized, information-based EA.
The Failings of Legacy EA Solutions
The EA practice has historically been dominated by centralized frameworks built on top of relational database engines. While these systems could capture and store vast amounts of architecture data, they struggled to generate insights or provide real-time visibility into an organization’s architecture. The nature of RDBMS makes it difficult to establish connections between business capabilities, IT systems, and processes, resulting in a lag in decision-making and strategic alignment.
A significant part of this issue also lies with the TOGAF model, a cornerstone of many EA practices. The lack of implementation guidelines for specific technologies left a gap to be filled by vendors whose software experience was based on SQL capabilities. TOGAF’s language and structure are rigid and overly complex, making it difficult for tools to adapt to disparate industries and business environments. The tools’ heavy reliance on relational tables means that much of the semantic richness that exists in organizational architecture is lost due to the inherent limitations of the underlying technology.
This critique of TOGAF is not unique to my observations. Svyatoslav Kotusev, a renowned academic and researcher in the field of Enterprise Architecture, has extensively studied the practical applications of EA frameworks. In his work, Kotusev argues that TOGAF and other popular EA frameworks often fail to deliver value in real-world scenarios due to their complexity and lack of practical guidance 2. His research suggests that successful EA practices often deviate significantly from the prescriptions of formal frameworks, instead adopting more flexible, situation-specific approaches.
So in practice, TOGAF’s complexity often bogs down EA initiatives, leading to either extensive documentation efforts with little practical value or limited implementations in domain and scope. As a result, EA often became synonymous with a bureaucratic exercise that offered little real value. When EA is not implemented well, the team becomes frustrated by lack of progress, and the organization’s “institutional immune system” may reject the program, and like a human immune system, this can have a lifetime memory3
The Rise of New EA Tools: Promises and Limitations
In response to the failings of legacy solutions and frameworks, newer tools based on property graph technology emerged with commercial success. These platforms successfully disrupted the EA tool market with their cloud-native, agile development approach, and lightweight metamodel. They excelled in areas like capabilities mapping and impact analysis, allowing for a more interconnected view of an organization’s architecture focusing on rapid time to insights and overall value. I am talking specifically of LeanIX 4.
LeanIX is a notable example of these new-generation tools, and SAP’s $1.2B acquisition of LeanIX was a strategic move, as it provided the perfect Enterprise Architecture tool designed to support large-scale cloud ERP migrations that SAP is pushing on their clients [5 6 7].
LeanIX provides easy-to-navigate visibility into both business and IT landscapes, enabling organizations to plan, execute, and optimize cloud migrations. As SAP pushes clients toward towards their cloud-based ERP systems like S/4HANA, a tool like LeanIX help would manage the complexities of such cloud transitions more effectively. By mapping the dependencies and interconnections between applications, businesses can reduce the risks associated with such large-scale move-to-cloud ERP deployments.
While this acquisition makes strategic sense for SAP and LeanIX, it took away a valuable partner for non-SAP organizations. LeanIX is still an active player in the EA space, but its primary focus may no longer be on the Enterprise as a whole.
Challenges for Large, Decentralized Organizations
Despite advancements in information architecture, thanks in particular to labeled property graph engines like Neo4J 8, significant challenges remain, particularly for large, decentralized organizations such as governments or higher education institutions (because academia is considered Organized Anarchy type-organization 9). I have personally witnessed extensive rework in the BA/EA space, single-purpose models, and duplication of effort. As far as I know, no tool currently has the capability to fully integrate and maintain a current-state architecture repository across disparate teams.
Don’t get me wrong, this is a daunting challenge! The duplication of work and the gaps that exist between business analysts, technical analysts, architects, operations, and support teams can be huge. While tools like LeanIX provide valuable insights, the dream of a decentralized, information-driven EA remains largely unrealized due to the lack of adequate tooling.
The Case for a Federated EA Model
The answer may lie in a federated model of Enterprise Architecture. Unlike centralized top-down approaches, a federated model would allow different parts of an organization to maintain autonomy over their own architectures while still adhering to a common framework. This approach could reduce rework, improve collaboration, and enable more dynamic, real-time decision-making.
I believe that Semantic Web technologies like RDF (Resource Description Framework) could provide the foundation for such a federated model. By allowing different data sources and models to be linked and reasoned over, RDF provides the flexibility and scalability needed to support decentralized EA across large, complex organizations. This is why I am advocating for the instantiation of ArchiMate, a leading EA modeling language, using RDF Ontology. It would not only enable better integration across teams but also provide a distributed, information-based approach to EA. (More of this on another post.)
It’s important to note that while the concept of a federated EA model might be promising, however I don’t have concrete examples of where this has been successfully implemented – or even attempted and failed. I need to do more research and experimentation in this area.
Frameworks like FEAF (Federal Enterprise Architecture Framework) and DoDAF (Department of Defense Architecture Framework) might shed some light on potential approaches to federated architectures in large, complex organizations. While these frameworks don’t inherently rely on ontology or semantic web standards, there have been efforts to extend them using ontological approaches for more flexibility and interoperability. Projects like data.gov aim to expose government data using linked data principles, which align closely with RDF technologies. Extensions to DoDAF have been explored where ontologies and RDF have been used to improve the expressiveness of the architecture and allow for better automation, validation, and integration. Need to read more about them.
Anyways. The technical aspects of implementing a federated EA model using Semantic Web technologies are complex and warrant a deeper discussion. On my next blog post, I will delve into the technical details of how RDF, OWL, and other semantic technologies could be leveraged to create a more flexible, scalable, and interoperable framework and the benefits of a semantically-rich, federated approach to EA.
Conclusion: Is a Federated EA Achievable?
While the idea of a federated EA model might hold promise, the reality is that no tool (or framework) has yet managed to fully realize this vision. Tools like LeanIX have made significant strides in supporting digital transformation efforts by providing better current-state and future-state insights, however, the challenge of maintaining a unified architecture across decentralized teams remains.
From my personal experience, I believe there is an opportunity here—not only to continue exploring this topic but also to potentially engage in academic research that addresses the gaps in current EA practices. The future of EA may very well depend on whether we can successfully move beyond the Ivory Tower and toward a more federated, information-driven model.
I would love to hear thoughts and feedback from the community. I’m not an expert, but this remains an ongoing area of interest for me.
References:
- Hohpe, G. (2020). The Software Architect Elevator: Redefining the Architect’s Role in the Digital Enterprise. O’Reilly Media. ↩︎
- Kotusev, S. (2018). TOGAF-based enterprise architecture practice: An exploratory case study. Communications of the Association for Information Systems, 43(1), 20. https://doi.org/10.17705/1CAIS.04320 ↩︎
- Jim Phelps (2020). Architecting the Architecture: Necessary Steps for Setting Up an EA Practice. https://er.educause.edu/articles/2020/10/architecting-the-architecture-necessary-steps-for-setting-up-an-ea-practice#fn3 ↩︎
- LeanIX. (n.d.). Enterprise Architecture Management. https://www.leanix.net/en/solutions/enterprise-architecture-management ↩︎
- SAP to Acquire LeanIX to Strengthen Its Business Transformation Portfolio. (2023, September 7). SAP News Center. https://news.sap.com/2023/09/sap-to-acquire-leanix/ ↩︎
- SAP NA President Lloyd Adams on Accelerating the Case for RISE, Cloud, and Business A (2024, April 14) I https://www.asug.com/insights/sap-na-president-lloyd-adams-on-accelerating-case-for-rise-cloud-and-business-ai#:~:text=And%20the%20lane%20that%20we,AI%20innovation%20in%20the%20cloud. ↩︎
- SAP Turns Up the Heat for On-Prem Customers to Move to the Cloud. (2023, July 25) https://accelerationeconomy.com/cloud/sap-turns-up-the-heat-for-on-prem-customers-to-move-to-the-cloud/#:~:text=Highlights,looking%20for%20more%20and%20more.%E2%80%9D ↩︎
- Neo4j. (n.d.). Graph Database & Analytics | Neo4j Graph Database Platform. https://neo4j.com/ ↩︎
- Robert Birnbaum (1988). How Colleges Work: The Cybernetics of Academic Organization and Leadership. Jossey-Bass, San Francisco, CA, USA https://eric.ed.gov/?id=ED301114 ↩︎
(Featured image generated by AI of Parmenides, widely recognized as the Greek philosopher who first defined ontology as a separate discipline distinct from theology.)
Alberto D. Mendoza 
Leave a comment