Relational the root of database strategy - The Impact of IBM's SAA Today; special editorial supplement - includes related article on data management

RELATIONAL THE ROOT OF DATABASE STRATEGY

At the nucleus of the enterprise-wide computing vision of IBM's SAA is the relational database. SAA's relational database family includes DB2 and SQL/DS for the System/370, OS/400 Database Manager for the AS/400 and OS/2 Extended Edition (OS/2EE) Database Manager for PCs. IBM's goal is distributed collection of interconnected databases.

As this SAA strategy evolves, software suppliers and user organizations, including IBM, are building products and systems on the three platforms, with the assurance that the common set of architectural features will expand into a consistent level of data access and management support.

Structured Query Language (SQL), the SAA database access language, is a key component of relational technology. It can be used as a query language or be embedded in programs written in Cobol, Fortran, C and PL/1. It also supports data management and data administration functions.

Today, much of the activity revolves around DB2 on System/370 mainframes and the OS/2EE DAtabase Manager. Three vendors who have committed to the SAA data management strategy are Sterling Software Inc., Canoga Park, Calif., Integral Systems, Inc., Walnut Creek, Calif., and Tesseract Corp., San Francisco.

Sterling Software is a long-time developer for IBM platforms. Its Answer series of fourth-generation tools lets users query mainframe databases from IBM and other vendors. Sterling's Micro Answer support extracting data from mainframe databases and downloading that data into PC-DOS products.

Scheduled for delivery by year-end, Presentation Answer, an SAA- compliant product, will support cooperative processing applications using the OS/2EE Database Manager and Presentation Manager.

Sterling committed to SAA over two years ago, noted Fred Braddock, vice president of R&D. "We got in very early, and the ground rules changed several times. This caused us to change directions, which was frustrating. But that's one of the penalties of being a pioneer."

Braddock notes that SAA is evolving, but the early components have settled down. For example, SQL is almost stable. The Common User Access (CUA) is less stable, with changes caused by shifts in hardware platforms.

However, Braddock believes the current set of functions are correct and necessary.

Communications is another issue. Although Braddock acknowledged that the CPI-C (Common Programming Interface, Communications) has a fairly well-defined set of protocols, he said that it is evolving with the "uncomfortable feeling that we haven't done the right thing yet," especially in the area of distributing data.

"There isn't a welath of experience or expertise there to know whether everything is connected right and working efficiently. There's not even a wealth of agreement on what distributed means, much less on how to do it," Braddock said.

Presentation answer will access data stored anywhere on a mainframe, download it to the PC and store it in the OS/2EE Database Manager as rows and tables. Then any product that can access the Database Manager can work with that data.

Sterling is writing interim Application Programming Interfaces (APIs) to expedite the communications links for Presentation Answer. One interface works between the application and CPI-C to pass SQL statements between the OS/2 and mainframe modules of Presentation Answer.

The first stage will take SQL and interpret it as required to access the mainframe data. Because the API uses SQL, any application that talks to SQL can talk to Presentation Answer. Thus, the second stage will be able to go to the mainframe or to the local database manager. "We will have the effect of joining both mainframe and local data," Braddock said.

According to Braddock, Sterling is architecting the additions so that as IBM components become available, they replace Sterling's. The additional effort is not too difficult or expensive, he noted, but it is an annoyance.

Integral, an IBM Business Partner, develops financail applications and human resources (HR) software. The company's products have been running in multiple database environments. Jeff Comport, SAA product manager, said Integral began developing DB2 products commensurate with IBM's annoucement of SAA.

Integral had been using upload and download techniques to combine DB2 data with DOS PC-based products, a risky territory where software can be made obsolete overnight. The SAA guidelines lowered the risk for Integral, he said.

"We see that as a big benefit of SAA. It allows us to involve ourselves in technologies such as workstations, communications and SQL databases both on the host and workstations," Comport said.

Integral has also embraced IBM's Cross Systems Product (CSP) as a development tool and many of the concepts of AD/Cycle as a development methodology. CSP has a strong affinity with DB2; one can exploit direct SQL access to DB2 without writing applications code. CSP can also read and write to the DB2 catalog, and has a high-level language for screen definition and data validation. Customers then can use CSP to modify Integral's products.

Integral supports DB2 analysis tools from Bachman Information Systems, Inc., Burlington, Mass. Thus, customers can make changes at the Bachman analysis level and generate optimized DB2 definitions that fit Integral's products.

Tesseract Corp. specializes in developing and delivering human resource management system (HRMS) solutions to large corporations. Committed to SAA, Tesseract uses the layered architecture, cooperative processing methods and graphical user interface (GUIe that are key elements of IBM's blueprint. Tesseract products have been available on DB2 since 1988.

The move to DB2 was market-driven. However, Tesseract's support of SAA came from the firm's interpretation of market direction.

"We were already in the process of evaluating how we could retain the requirements for sophisticated applications, requirements such as the need for central data and control from a central location, and still take advantage of the emerging workstation," said Mary Kay Marvin, vice president of marketing.

Using Windows 2 from Microsoft Corp., Redmond, Wash., Tesseract developed a prototype illustrating its concept of how an SAA application would look and act. Demonstrations to both customers and IBM confirmed SAA as a strategic direction.

HRMS Intuition is Tesseract's first product developed under SAA guidelines. Now in beta test, Intuition is a cooperative processing application, connecting the mainframe HR DB2 data with the OS/2EE user at a PS/2 workstation.

"SAA allowed us to focus on building our products on a single platform and concentrate on our applications rather than spending resources on developing platforms," stated Gary Durbin, executive vice president at Tesseract. "In addition, it has focused our clients on relational database."

Tesseract feels that IBM's tools and pieces are not allng to Durbin. Missing elements, he said, include communications, a consistent implementation of SQL, and query and other relational database tools.

Tesseract is plugging the gaps by building tools its customers demand. For example, the firm has developed utilities to support replica updata of its databases. To build tools that must later be pulled out in favor of IBM's is an expensive but necessary aspect of a commitment to SAA, Durbin indicated.

While application vendors see the value of SAA, user reaction varies from enthusiasm to skepticism. A firm's first concern is the software's ability to meet business needs. SAA compliance remains a technical concern. Companies are testing the SAA waters by purchasing and installing SAA-compliant products.

"DB2 is our carrot into the user's thinking," said Tesseract's Marvin. Clients and prospects want production systems that support DB2, he said. "When we talk about SAA, we say that DB2 is our first step . . . and now let's look at additional phases."

Chevron Corp. of San Francisco is running Tesseracths IMS HRMS products and beta testing Intuition. It will migrate to the DB/2 version early next year.

Jay F. Stright Jr., manager of Chevron's HR information system, said Chevron did not have an HR function until 1985. "We were very late geting started, so we didn't have any baggage to drag with us.

"A fundamental requirement of a [HR] system is that it must facilitate change," he said. Stright said that a company, as it grows, needs to know where it has been and where it wants to go. "Thus, a primary requirement for our [HR] system was the time relational model. That led us to Tesseract."

Stright said the Tesseract architecture, unlike traditional systems, allows users to design and make changes themselves.

Although Chevron had DB2 running in a test mode, the HR application will be the first significant DB2 application. "We are so committed to DB2 that we are joint-funding DB2 enhancements with other Tesseract users," said Stright.

Chevron has not had any problems in installing and testing the database, according to Stright. The major concerns are peripheral to the database. Stright feels that DB2 is better positioned than IMS was at this point in its development. With Intuition, users do not have to learn the details of SQL.

George Rogers, information systems consultant at Puget Sound Power and Light, Puget Sound, Wash., is the technical project manager for installing Integral's payroll, personnel and time recording system.

SAA compliance was a factor in choosing the Integral system, but "first we have a business need to satisfy," Rogers said. "Then SAA came into play in looking at the technical practicality with which we satisfy that requirement, and also from a durability standpoint."

Puget Sound P&L felt that DB2 would offer more flexibility and longevity. The Integral products' use of CSP was also important.

Rogers said there is a significant learning curve in converting to DB2, but "it is easier than developing a Vsam structure. There is more flexibility, more opportunity to alter decisions without having to restructure the data desig."

Rogers said he has not experienced any significant problems. "We like DB2. We like the flexibility and are quite confident that we can keep it tuned for our applications." His one concern is performance. However, since the company has been workign with DB2, performance is continuously improving.

Vendors commiting to SAA are conscious of the risks. Bill Leckonby, president of Tesseract, said, "This is an investment, and an investment for a return that we don't know when is going to come."

SAA, with its emphasis on the relational model and cooperative processing, is changing the way applications are developed. It is encouraging a more careful architecting of applications that will be advantageous to both vendor and user organizations in the long run--despite the fate of SAA.

Right now, Comport said, "Vendors have made SAA more of an issue in the selection process than it would be from just the customer point of view."

However, he said that functional requirements are increasingly demanding capabilities that can only be delivered by SAA technology.

Rogers said, "From a user perspective, we really don't have a need for SAA tools yet, but these will come as vendors get their products to market."

GARTNER ANALYSIS:

DATA MANAGEMENT

Data management constitutes much of the foundation of SAA (the other main component being CUA on programable workstations). In fact, it makes sense to think of the entire SAA Common Programming Interface (CPI) as extending the notions and functions provided by the four relational database managers--DB2, SQL/DS, OS/2EE Database Manager and SQL/400.

SAA data management delivers two promises--application portability support and a means of enabling a "single-system image."

Application portability is a difficult and complex concept, and involves many facilities besides database management. But database management, through the SQL programming interface, can account for over half the work involved in portability. If IBM solves the problem of database programming portability, then its customers will have moved halfway toward application portability.

However, one must look deeper than application portability to discover a key business impact of SAA database management: programmer training and productivity. While application portability per se is not a widespread user requirement, every IBM customer is concerned about programming costs. By (making significant progress toward) uniformizing the programming interface for much installation data, IBM is contributing toward containment of those costs.

Nonrelational data (stored in IMS/DB and Vsan files) falls outside this realm. The existence of such data will dilute the impact of cost containment.

The main business impact of SAA data management will result from the single-system image thrust--to an end-user of a programmable workstation, the location of (relational) data will not be a concern. SAA facilities will retrieve and update data automatically in response to a user's commands. That data could be located anywhere.

What will be the business impact of this database-level connectivity? IT can amplify the effectiveness of the existing corporate software inventory. It may also enable business to better exchange data with clients.

In other words, by extending the reach of the software already in place, this "single-system image" may increase the return on the total dollars invested in electronic equipment.

Of course, for all these benefits, there is the question of whether IBM's database management strategy needed SAA. Many observers feel that IBM would have delivered distributed database capabilities without SAA. Nevertheless, SAA gets "credit" for the benefits of distributed database technology.

Rich is president of PFS, Inc., El Segundo, Calif., a consulting firm specializing in the design and deployment of end-user-oriented database systems.

COPYRIGHT 1990 Wiesner Publications, Inc.
COPYRIGHT 2004 Gale Group