- What is IT Infrastructure?
- The Philosophy of Adaptive Infrastructure
- Benefits of Adaptive Infrastructure
- Is Your Organization Ready?
- Summary
This chapter is from the book
This chapter is from the book
This chapter is from the book
Benefits of Adaptive Infrastructure
So far, most of the benefits of adaptive infrastructure have been fairly obvious. Other benefits may not be so apparent. Some of the less understood benefits are detailed below.
Developing an Adaptive Range
Understanding exactly what is "adaptive" about adaptive infrastructure can be a complex task. A number of dimensions are involved in quantifying the flexibility or adaptive range of infrastructure, as shown in Figure 1.7.
){kind=link}
Figure 1.7 Measuring Adaptive Range
One obvious measurement is cost. Everyone wants a fixed, low cost, but how you deliver it is the real problem. Speed is the next measurement that attracts the most attention. Speed refers to the timeliness with which IT can respond to users and implement the business processes they want. However, speed is a derivative measurement. Certainly, buying more powerful hardware can increase the transaction rate, but infrastructure isn't equipped with a speed knob that can be turned up or down as needs dictate. To increase speed, other dimensions must be addressed.
Scalability.
You can build in some headroom (adaptive range), so that boxes and boards don't have to be swapped when increasing the number of users for an application. That is a relatively easy (though expensive) way to be adaptive, but it isn't sufficient to accommodate the rapid changes in business processes that users are creating.
Presentation.
Another dimension that affects virtually every IT organization is presentation: the way information is presented to users or business partners. Historically, the presentation layer has shown little adaptiveness. Organizations now spend a great amount of time and effort converting more traditional 2-tier, client/server applications into Web-based applications. Unfortunately, designing Web-only presentation solutions is equally limiting. Other presentation methods, such as personal digital assistants (PDAs) and interactive voice response (IVR), can't support a full-page Web display and will require completely new development efforts, which are expensive and time-consuming.
Presentation independence alone doesn't guarantee sufficient adaptability, however. Infrastructure planners should be very focused on the front end, but if the application-to-application integration issues aren't addressed simultaneously, the result is taller "stovepipes."
Partitioned complexity.
The ability to partition functionality and complexity within the infrastructure is another measure of adaptability. If the infrastructure cannot be effectively partitioned, the resulting complexity will spread throughout the organization and eventually become unmanageable. You must effectively manage the interfaces between applications and between infrastructure components, both for the enterprise and for applications used by external partners.
Integration/reuse.
Infrastructure integration and reuse are also measures of adaptability. The typical organization requires a dramatic increase in the reuse of infrastructure code, other technology components, and skills to increase adaptability and speed of deployment. Reusable code is the opposite of legacy code. Whereas legacy code can be difficult to maintain, enhance, and integrate, the most adaptive code can be adjusted as the business evolves.
These factors and more contribute to the adaptability of an organization's infrastructure. This book discusses many of these issues in greater depth.
Increasingly, you must describe the value proposition for infrastructure directly to the business users in terms of the discrete service levels that they want delivered. By turning to a discussion of service levels, you can influence business users to consider more than just the immediate impact of a single application. You must convince the business to consider the value of adaptability, because it will take more money, effort, and time to deliver than single implementations.
The Importance of Reusability
Of all the concepts discussed so far, reusability is the most crucial. The patterns, platforms, and models used to create an adaptive infrastructure all depend on reusability. Without an effective approach to this issue, it will be difficult to apply the additional rigors of specifying pattern components and creating predictive cost models.
Reuse isn't simply a matter of being thrifty or making do with whatever hardware happens to be lying around the shop. In fact, reuse often involves just the opposite~weaning business units away from archaic legacy systems that are no longer cost-effective and replacing flawed applications or integrating them with other applications within the enterprise.
When implementing a reuse policy, the biggest challenge is getting people used to the new policy and getting them to fund it appropriately. It's difficult to tell business executives that their infrastructure requirements are physically impossible, or that they can't have an application because it's too difficult to support.
Many business units have the financial and human resources to subsidize non-standard implementations, but they don't. Instead, the IT department ends up absorbing the extra expense to support the deviation, or devoting extra resources and effort to making sure it works properly. This will continue until you present errant business units with a reasonable alternative or a bill for the extra time and effort. If the business units only encounter half-hearted resistance from IT, you will probably spend most of your time doing ad-hoc support, and you can forget about creating an adaptive infrastructure.
Reuse policies are both a logical precursor to and an integral part of implementing adaptive infrastructure techniques. Designing reusability well requires that you leverage all the concepts introduced here, particularly patterns and services. Once your organization is ready for reusability, or actively practicing it, you are ready to implement an adaptive infrastructure.
Reusable Components
The following table is a list of the infrastructure components that should be included in any reuse efforts.
Reuse and Adaptive Infrastructure
The concept of reuse has been around since the arrival of object-oriented programming (OOP) languages in the 1980s. OOP promised that developers could write applications, then reuse large blocks of code, called "objects," in other new applications.
OOP failed for a number of reasons. The concept was too complex and fine-grained for the average programmer to grasp. In many cases, software developers created objects that were so specific to a particular application that it was difficult to reuse them without a major rewrite.
As a result, developers were forced to start from scratch each time. Still, the OOP concept was compelling enough that others have since attempted to introduce reuse at higher levels of abstraction, such as security, middleware, and networking.
Scalability was another problem. An organization with a 100-percent object-oriented approach had to support millions of objects. This proliferation resulted in a separate object for every customer in a particular database. The history of some of object-oriented database wars shows this simply doesn't work, as shown in Figure 1.8.
){kind=link}
Figure 1.8 What Kind of Reuse Is Achievable?
Table 1.1 Reusable Components
|
Basic Components |
Metacomponents (Define standards for architecture, interoperability, etc.) |
|
Network
|
Business rules |
|
Hardware
|
Process models |
|
Software
|
Documentation templates
|
|
|
Action diagrams |
|
|
Data models
|
|
|
Algorithms |
|
|
Plans
|
NOTE
A more extensive infrastructure component list can be found in the component catalog listing detailed in Chapter 2.
In fact, reuse is best achieved at the infrastructure level, not within applications but between applications. For this reason, you should focus reuse efforts on broader external components.
A banking and financial services firm, for example, might have multiple credit authorization processes to support various business units. To promote reuse, you could consolidate these various applications into a single credit authorization module. It doesn't matter whether this common module is object-oriented or not. Reusability stems from the fact that different applications share it.
Third-party software vendors are leading the way in this area. SAP, for example, includes integration components in its architecture in the form of Business APIs (BAPIs). PeopleSoft leverages middleware from BEA Systems to integrate various modules, processes, and data types within its application suite. These types of components and middleware are crucial for organizations that are integrating existing applications into e-Commerce and supply chain solutions.
Applications that present IT services to external business partners depend heavily on component-oriented solutions. Each part of the supply chain becomes a component with interfaces exposed to its neighboring component, permitting the development of some fairly complex supply chain configurations.
The key to success in infrastructure planning is to change the way things are done so that applications become easier, cheaper, and quicker to integrate. Success also means being able to run these services long-term with high quality.
Another form of reuse that emerged in the 1990s involves placing a "wrapper" around legacy applications using interface definition language (IDL) models such as CORBA or DCOM. This practice is a perfect example of internal applications not needing to be object oriented. Typical mainframe applications can be migrated into a component framework by wrapping them inside an IDL.
Today, approaches such as Java 2 Enterprise Edition (J2EE) and Microsoft .NET are moving further down that path. These approaches provide not only "wrapper" capabilities, but also the beginnings of a set of Web services that can be leveraged across applications, processes, data types, and businesses. Web services represent the latest opportunities for leveraging and reuse at the intersection of the application and infrastructure worlds. These approaches take advantage of standards such as Extensible Markup Language (XML); Universal Description, Discovery, and Integration (UDDI); Simple Object Access Protocol (SOAP); and Web Services Description Language (WSDL).
In summary, successful reuse efforts don't get deeply involved with fine-grained componentization. Rather, these broad efforts emphasize in-tegration components and focus on the infrastructure level.