The Art of Enterprise Information Architecture: A Conceptual and Logical View

This chapter is from the book

Art of Enterprise Information Architecture, The: A Systems-Based Approach for Unlocking Business Insight

Learn More Buy

This chapter is from the book

This chapter is from the book 

Art of Enterprise Information Architecture, The: A Systems-Based Approach for Unlocking Business Insight

Learn More Buy

4.3 Architecture Principles for the EIA

Architecture principles are a set of logically consistent and easily understood guidelines that direct the design and engineering of IT solutions and services in the enterprise. These principles provide an outline of the tasks, resources, and potential costs to the business for their implementation, and they also provide valuable input that can be used to justify why certain decisions have to be made.

Architecture principles should enable the EIA Reference Architecture as a tool that can help you map between the organization’s business goals, business architecture, IT landscape, and the solution delivery, and help the Enterprise Information Architecture to be the conduit to understand the implications of planned new business requirements.

An EIA should be based around many of the principles in Table 4.1. Without such guiding principles it is likely that any solution will become fragmented or it becomes increasingly difficult to exploit design elements across the enterprise. The architecture principles for EIA shown in Table 4.1 are by their nature generic and would need refining, shaping, and probably adding to in order to usefully serve on any specific instantiation of the EIA Reference Architecture. The column labeled “New” indicates if the architecture principle is a new one due to recent changes in existing or the appearance of new business requirements. The “Domain” column indicates the data domain where the architecture principle is applicable the most. Note that this does not imply that the principle is completely irrelevant in another domain; however, we tried to distinguish strong relevance from minor relevance or irrelevancy in the Domain column. The note of “All” in the Domain column indicates that the architecture principle applies to all data domains. Following the table, each architecture principle is explained in more detail.

Table 4.1  Architecture Principles for the EIA

1. Deploy enterprise-wide Metadata strategies and technologies. Metadata management provides unambiguous definition of data and its history (how it is transformed and manipulated throughout the organization) to enable unified information integration. A comprehensive Metadata management solution improves data quality by providing a full understanding of the data. It is essential to have consistent Metadata enabled to easily build a set of repeatable and reusable IT processes because it centrally and completely documents data and applications. It enables business users and technical developers alike to have a common understanding of the data assets already available to the business and a precise meaning for that data and its usage.

2. Exploit analytics to the finest levels of granularity. The value of data is requiring that ever deeper analytics are exploited to understand smaller events that may influence specific groupings or segments of information such as customers, suppliers or products. This requires not only data access, but that the tooling can crawl through vast quantities of information to extract significant (potentially small but high value) events in a manner that is easily consumed by end users and business processes alike. Techniques such as data mining with visually rich interfaces to understand statistically significant events is one opportunity and being able to slice and dice through OLAP (On-line Analytical Processing) tooling to fine grained data is another. Reporting tools can also be combined with data mining tools to enable mining of data by less qualified end users who see the solution as nothing more than another report that returns a result set (for example scoring a set of customers). Remember that once data has been summarized, that lower level of detail is lost, and if it should prove useful at a later date, may not be retrievable.

3. Exploit Real Time and Predictive Analytics for business optimization. This principle requires new analytical capabilities to be able to derive insight into information in real time when the event occurs. Complex event processing capabilities for reducing loss in water and electricity networks providing real time insight into consumer demand are one example for this real time analytical architecture principle. Predictive Analytics using various algorithms enable an organization to foresee events or values of attributes in the future with a certain probability.

4. Enable KPI-based BPM. In a globally competitive marketplace, the business strategy must be measurable—otherwise it cannot be controlled. This architecture principle thus requires a holistic approach in the EIA to allow the definition, monitoring and reporting on critical business KPIs so that the executive management team can manage an enterprise based on business performance on an ongoing basis. Note that the KPIs themselves could be considered Business Metadata and thus require Metadata management lifecycle capabilities.

5. De-couple data from applications enabling the creation of trusted information which can be shared across business processes in a timely manner. There is a need for a consolidated, accurate, consistent and timely view of the core business entities using a common model for each entity. Without such a view it will be very difficult to define a set of services that can adhere to some common meta-model to drive transformations for data that is held in messages, Master Data solutions, DWs, or Operational Data stores. This view needs to take into account the business definition of data, as well as the physical and logical definitions for that data in any model developed. Difficulties arise if there is no common model whenever there is a need to create and extend enterprise wide business processes, to create consistent reporting, and to re-use services.

In some data domains, such as Master Data, the information is extracted from a variety of sources, harmonized and loaded into a centralized system, managing it and making it available to consumers through services decoupling the data from the consuming applications. Timely and trusted information includes data governing rules that define availability, standardization, quality and integrity. Information requires also associated Metadata describing its source, quality and other relevant attributes so that it can be trusted. This principle is particularly applicable for the Master Data domain because Master Data has the characteristic of high re-use across a large number of business processes. However, there might be applications with too many embedded business rules so that the de-coupling of the data would be cost-prohibitive or even impossible. In such a case it might be appropriate to consider the application itself as information source and consume it through a Cloud Service for example.

6. Strive to deploy an enterprise-wide search capability. One consistent search engine across content repositories, databases, applications, collaborative environments and portals to shorten the time to identify useful information, is becoming a constant requirement for business users. By being able to access all such sources through one simple interface, the job of quickly identifying and making use of the best data for a particular process or job is much simplified. It should not matter whether information is on an intranet page, within a content management system, buried deep inside an ERP application or CRM solution, residing within a legacy database or even in an e-mail system, the search engine deployed should be able to find it. Another dimension of enterprise-wide search is the ability to extend the intranet search to resources such as the Internet in a unified manner.

7. Compliance with all Information Security requirements. This principle includes accounting for several layers of security, identification of a broader risk analysis strategy and the definition of specific rules around Information Security. Basic capabilities are proper authentication and authorization mechanisms which are required by this principle.

8. Compliance with all relevant regulations and Information Privacy legislation. All information assets should be protected regarding information legislation requirements. Also, all information assets must be managed in compliance with all legal regulations such as Sarbanes-Oxley.¹³

9. Deliver de-coupled, trusted information through Information as a Service (IaaS) so that information services in a SOA are reusable and shareable services for the business like other business services. Basically this architecture principle complements the fifth architecture principle. Once information has been decoupled from an application and is trustworthy, it makes sense to deliver it through Information as a Service. Thus, this principle is one of the core components of a well-formed SOA strategy and is part of the definition of the IOD approach to deliver information in context at the right time to the right application or business process. Implementing highly modular, loosely coupled systems and services is the most efficient way of taking advantage of reusable services and minimizing the cost associated with the duplication of processing tasks. This principle also facilitates leveraging services that are provided by other parties. This means that information is packaged as a service to business processes, so that consistent, manageable information is made available to every process in a standardized way that enables re-use and business flexibility. Deploying information services makes them also discoverable in the same way business services are discoverable through a Service Registry and Repository (SRR). (More details on the relationship between an EIA and SOA can be found in Chapter 2.)

10. Deploy new levels of information lifecycle management creating actionable information. This architecture principle requires managing all information assets across their entire lifecycle efficiently. In addition, this principle mandates to create actionable information by emitting notifications and events if information changes and the new values satisfy certain pre-defined conditions and rules. Actionable information can be categorized such as business events (for example, a bank might place a rule that two months before a fixed term deposit expires, a customer care representative must be notified via e-mail to contact the customer), infrastructure events (for example, assignment of a value to each information asset in order to manage storage costs efficiently), and regulatory events (for example, the compliance with legal requirements to retain or delete an information asset in a timely manner such as e-mails related to a certain law suit at court).

11. Apply Cloud Computing delivery model to Information Services. Once information is available through Information Services as indicated with the ninth architecture principle, the Cloud Computing delivery model can be applied to them. With this architecture principle, design consideration must be made for the Operational, Unstructured and Analytical Data domain if these services can be deployed into Cloud Computing environments reducing cost and further improving flexibility. Pushing an SRR into a Cloud Computing environment externally hosted exposes the whole internal SOA infrastructure because now service discovery and service routing is available if and only if the external cloud service provider is available.

12. Improve cost efficiency of the IT infrastructure, possibly now also using Green IT techniques. The EIA is typically not deployed on the green field. A guiding principle for each phase of an iterative rollout of EIA is the improvement of cost efficiency. We consider cost efficiency in this case in three dimensions:

• A Green IT perspective. Energy costs are, for many data centers, the major part of the operational costs. Thus, selecting appropriate hardware platforms as part of this architecture principle that are enabled for Green IT consumes less energy and, hence, reduces costs for the IT infrastructure.¹⁴ Achieving this goal requires, first, analyzing and measuring energy efficiency in the data center with static and dynamic thermal measurements. Once the current situation is assessed, techniques in energy efficient system design such as modular systems and the use of new energy efficient systems can be applied. This can be complemented by using virtualization techniques to consolidate or to reduce complexity leading to decreased energy consumption.¹⁵ Further optimization can be achieved using advanced energy management.
• Cost efficiency can be improved by applying advanced IT service management principles. In this category belongs the integration of workload management and traditional facility management in order to manage energy consumption by deploying different measures such as sensor networks in the data center. Another step in the area of advanced IT service management is to reduce complexity in administration thus reducing management overhead.
• Particularly relevant for information aspects is to apply DW and federation principles to align IT and business oriented service management disciplines. Furthermore, if the definition of cost efficiency is interpreted in a broader fashion, this principle demands exploitation of information management capabilities to reduce loss in energy and water networks making these utility networks smarter and therefore friendlier from an environmental perspective. As a side effect, it decreases costs operating these networks by avoiding energy waste, for example. Another scenario for this broader understanding of the cost efficiency principle is the creation of smart traffic systems based on information management capabilities reducing carbon dioxide emissions through reduction of traffic jams. This reduces cost from a perspective of not having the need to reschedule meetings if attendees are stuck in traffic causing decision or project delay, for example.

13. Deliver information with appropriate data quality. Data quality is the cornerstone of decision making—without high quality data results can always be challenged across differing business areas. This leads to impaired decision making, poor coordination across the business and ongoing costs to clean data in a piece meal or siloed fashion. Note that data should be classified in terms of its value and hence its requirement for absolute accuracy, for example, enterprise Master Data is often regarded as highly valuable data because it is used to enable a company to present itself most favorably in all customer interactions. As such it must be highly accurate and is normally considered more valuable than data that is only used within one LOB.

14. End-to-end inter- and cross-enterprise information integration. This principle requires the deployment of comprehensive end-to-end Enterprise Information Integration (EII) capabilities to seamlessly support enterprise information integration initiatives such as Master Data Management, enterprise-wide BI solutions or information integration for cross-enterprise solutions such as an RFID-based Track and Trace solution for e-Pedigree¹⁶ compliance.

15. Develop an EII strategy with optimization of data transport, federation, and placement. This architecture principle expands the previous one regarding an optimization aspect. As well as managing all the data stored in various repositories around the enterprise there is also a pressing need to manage data that is moving around the enterprise at any one time. These data may take the form of flat files being managed in batch, message queues, XML files or replicated data. The alternative to moving data is leaving data in place using federation to access it. This approach tries to ensure that multiple copies of the same data are not proliferated around an organization, being maintained multiple times and potentially having its value reduced if there are multiple update points to it which might result in multiple versions of the truth. Federation might not always be an option for performance reasons, in which case a data movement strategy to ensure data is kept in sync between copies is preferable. By developing a comprehensive strategy to address the need for unified information integration solutions using data movement and federation techniques as appropriate enables the business to be certain that data is consumed in the most effective manner at the right point in a business. This principle is closely tied to the Metadata principle: Without consistent Metadata and governance of that Metadata it becomes very difficult to maintain a company wide approach to movement and placement of data. This can result in fragmented data feeds, the same data being managed in the infrastructure many times in an inconsistent manner, resulting in excessive re-development, costly change, and poor data quality.

16. Virtualize information whenever possible. Storage virtualization and file system virtualization are the basic layers to implement this architecture principle. In Chapter 6 detailing the Operational Model, you can find operational patterns helping to implement this principle.

17. Deliver operational reliability and serviceability to meet business SLA to ensure access to Structured and Unstructured Data at all times. Much of a company’s digitized information is unstructured including rich media streaming, website content, facsimiles, and computer output. Unlimited access to such information alongside relational based information is critical to enable a complete view of the entire information available to solve a business need. Today’s business solutions require that all company information is available to its employees where necessary. It is becoming increasingly common to see Structured and Unstructured Data residing alongside each other within a set of services that a particular business process needs.

18. EIA should reduce complexity and redundancy and enable re-use. Implementing highly modular, loosely coupled systems and services is the most efficient way of taking advantage of re-usable services and minimizing the cost associated with the duplication of processing tasks. Additionally, this principle also facilitates leveraging services that are provided by other parties.

19. EIA should be based on open standards. This will enable the use of multiple technologies for the interoperability of the systems in the enterprise and external partners. Open standards facilitate interoperability and data exchange among different products or services and are intended for widespread adaptation.

20. Enterprise information assets must have a business owner and be part of end-to-end Information Governance. An Information Steward is responsible for defining the rules for information usage. This eliminates confusion regarding who can maintain, manage and change that information. Information must be viewed as a business asset so it should be managed accordingly to ensure that its value to the organization is maximized.

21. Align IT solution with business. Organizations articulate a set of priorities that serve as a guide where to focus the enterprise efforts.

22. Maximize agility and flexibility of IT assets. This principle requires the IT Architect to consider agility and flexibility aspects of the EIA. For example, the federation techniques mentioned earlier might be used to more loosely couple applications and information sources.