DBMS, April 1997
Leveraging Chaos
A New Infrastructure to Help You Transition to a New Computing Environment.
For IT organizations, the most frustrating aspect of moving to distributed computing is trying to figure out how to move forward effectively while keeping existing lines of business applications running. In the past, any time organizations thought about wiping the slate clean and building a brand-new architecture populated by only well-architected applications, they usually gave up. In my view, the best approach for IT organizations is to leverage existing software that provides true business value and to move forward incrementally. I call this approach the Hurwitz Hyper-Tier model.
When I first began to analyze the major transition from the mainframe as the focal point of corporate computing, it became apparent that a new infrastructure would have to emerge to support a new computing environment. This transition to a new mode of computing has intersected with a new way of doing business -- one where tight interaction between customers and suppliers becomes the norm. By now, everyone has heard of the virtual corporation, a forward-moving view of where businesses will be in the next decade. However, many organizations are confused about how to reach this mecca of technological innovation and increased productivity. The amount of change necessary to an organization's existing infrastructure can seem overwhelming.
A Network-centric Infrastructure
The typical IT software environment is chaotic. Most computing environments consist of a hodgepodge of systems, applications, data, and chaos. Attempting to hardwire applications to resemble client/server computing simply doesn't work. This model leverages the chaos that exists within an organization by allowing the multitude of systems, applications, and user interfaces to work in harmony.
How is this accomplished? One of the basic tenets of the model is a network-centric infrastructure. The model connects users, applications, services, data, and business rules in one vast patchwork of networks. (See Figure 1.) These networks are flexible and open, allowing a variety of clients, servers, and data sources to interact. The keys to this infrastructure are the protocols that enable ubiquitous communications including TCP/IP, HTML, and HTTP.
A Layered Approach is Key
All of the technology in this network-centric infrastructure fits into three layers. These layers provide a logical model for viewing and managing existing, chaotic business logic, rules, and data in an organized way and migrating them toward a more flexible structure. (See Figure 2.)
Access Layer
An access layer provides client-side software and includes everything from terminals to client/server fat clients, plus Web browsers, "information appliances," and some Internet-oriented dedicated clients. For a user, the access layer is the appropriate interface to applications. The goal with this layer is to determine the best way for your users to complete their tasks. Often an organization has many types of users, each with different needs. Creating an independent graphical user interface will let your organization utilize a number of access options within the same system. Many organizations are beginning to achieve quick results with the access layer by giving users Web browsers. This added functionality will satisfy users' basic needs, and the IT organization concentrates on the two layers tied closely with your business: the coordination and resources layers.
Coordination Layer
The coordination layer connects and manages client interactions (access layer) with system resources, and it includes Web servers, directory services, security, and various types of middleware (for example, to manage transactions and messages). It acts as both a traffic cop and translator between the other two layers. To maintain maximum flexibility, the various services within the coordination layer are separate from the applications and data in the resource layer. A Web server adds the hyperlinking capabilities and the ability to jump from place to place without preprogramming. Technologies such as CORBA, Microsoft Corp.'s Transaction Server, BEA Systems Inc.'s Tuxedo, IBM Corp.'s Transarc middleware services, and Tibco Inc.'s Enterprise Transaction Express all are examples of the middleware that exist the coordination layer.
Resource Layer
The resource layer consists of the data, including databases, entire applications, parts of applications, business rules, metadata, and components. Components are business rules, application functionality, data, or resources that are encapsulated so they can be used by more than one application.
The ultimate goal of the resource layer is to combine modular, component-based resources combined clean data and meta definitions. Most organizations, however, look nothing like this today, and the idea of achieving a unified data source may seem overwhelming. Don't let it be: Take an incremental approach. Look at your data. Is it inconsistent? Are there six different definitions for "customer"? If the answer is yes, organize a task force to work on standard data definitions. If you have hundreds or thousands of different data points, start with 15 or 20 key definitions and work incrementally toward cleaning up all of your data. The closer you can get to standardization, the better for both your systems and their users.
Each layer in this model contains many computing resources. For an end user, the access layer could be a simple, thin-client interface with a touch pad, a fat-client interface with a fully configured PC, or a Web browser. What the end user wants is access to data or logic in the resource layer.
The coordination layer provides the middleware that enables the access and resource layers to communicate. What happens is that a user invokes an action (or request), and the action uses an element of the coordination layer via the most appropriate form of middleware to access the appropriate rules, logic, or data. Layers are independent of each other to maximize flexibility. Some coordination layer solutions to consider include Web servers, transaction monitors, and directory services.
Note that the layers in this model are not the same as those found in a typical three-tier client/server model. There is no single service with a single piece of middleware, no single graphical user interface, and no single type of data all tightly connected. You should think of each layer as a category -- an organizing principle. This layer approach will help you plan the function of your various pieces and determine what to work on first.
The Internet as a Catalyst
The model builds on the flexibility inherent in the Internet protocols and extends the idea of hyperlinking to encompass existing legacy and client/server systems. Thus the Internet has been a key catalyst of this computing model.
Achieving Incremental Change
Break apart the pieces of your system and evaluate each piece from a business standpoint and a systems view. As yourself: Is this piece essential to my business? If yes, how can I modularize it to make it more flexible and scalable? If no, what can I choose to make it more modular? Breaking down your environment into digestible chunks will let you determine exactly which areas need your attention immediately and which pieces can afford to wait.
And finally, in the age of environmental consciousness, apply the practice of recycling to your systems as well. Look at all of your existing applications and application services and ask if they can be applied to any other area of the system. Many times, these components can be "recycled," saving both precious time and money.
Judith Hurwitz is president and CEO of Hurwitz Group Inc., a technology and management consulting company based in Newton, Massachusetts. Hurwitz Group focuses on the business impact, use, and deployment of distributed technology. You can email Judith at jhurwitz@hurwitz.com or visit her company Web site at www.hurwitz.com.
Figure 1.

--One of the basic tenets of the model is a network-centric infrastructure. The model connects users, applications, services, data, and business rules in one vast patchwork of flexible and open networks.
Figure 2.

--The three layers of the model: the access layer, the coordination layer, and the resource layer.
Subscribe to DBMS and Internet Systems -- It's free for qualified readers in the United States
April 1997 Table of Contents | Other Contents | Article Index | Search | Site Index | Home
DBMS and Internet Systems (http://www.dbmsmag.com)
Copyright © 1997 Miller Freeman, Inc. ALL RIGHTS RESERVED
Redistribution without permission is prohibited.
Please send questions or comments to dbms@mfi.com
Updated Tuesday, March 18, 1997