The next generation of software for distributors

BY CLARK YENNIE

Special to The Wholesaler

In the 1970s the Distronics time sharing system, with its tiny price matrix and hard copy terminal, was state-of-the-art in first-generation distribution software. The second generation brought the computer in house, with software like shims, which debuted in the late 1970s. Shims quickly spread to more than 500 supply houses and is still in use today at around 200 of them. Shims has survived for almost 30 years because it has an easy-to-modify source code.

Another major breakthrough in distribution software came with the Eclipse system, which replaced dumb terminals with pc-based workstations and forecasting based on summary data with forecasting based on transaction details. The pc workstations were so much faster than dumb terminals that users were able to increase productivity dramatically, and the system’s forecasting and automated purchasing reduced inventories dramatically.

Many incremental improvements have been made to distribution systems over the ensuing years. But one thing that has changed dramatically in the past 15 years is computer hardware. Computers are a thousand times faster and have a thousand times greater storage capacity today than they had 15 years ago. Software tends to lag behind hardware, but the hardware has changed so dramatically that software breakthroughs are now long overdue. New software will be released this year that will rectify that situation.

To appreciate how much of a difference the hardware makes, consider the evolution of database management systems. The first prototype relational database management systems were developed at ibm and Berkeley in the 1970s. It took teams of researchers several years to get these systems to work at all, and then only in the laboratory. It wasn’t until the 1980s that Oracle released the first commercial rdbms, and it had to struggle for years to make its product stable. Today, graduate students routinely create database management systems as thesis projects.

What changed? The hardware, mostly. Those early pioneers had to struggle with very slow machines with very little memory. With a modern computer you can store an entire database in a program variable.  Much greater speed and capacity make life simpler for programmers.

The next generation of software for distributors will dispense with the conventional database management system. Instead, it will have its own built-in database system, and its database will double as a data warehouse. It will be time addressable. The whole system will be tightly integrated to make it faster, simpler and more portable. The database will also be auditable, because it won’t overwrite previous versions of records when it is updated. Instead, it will append the new versions to the file and index all the versions by time and date. To audit all of the changes to a record, an application can simply walk back through all its previous versions.

When generating reports, the user will specify a time address and the system will present the data as it appeared at that moment in time. The data will be exported to the user’s workstation, where it will be imported into a spreadsheet or perhaps a conventional relational database for further analysis. Thus the database as we know it will move from the server to the desktop and the data warehouse will merge with the application on the server.

Next-generation software

In the next generation:

• Data will be stored in Unicode, so all international character sets will be supported. Forms and menus will be translated on the fly into the preferred language for each user.
• Transactions will not be shredded as they are today. Instead, they will be stored intact and indexed so that the data will accessible in a variety of formats.
• The user interface will be graphical, of course, but it will be more efficient than current gui interfaces. Don’t forget, it took years for designers to refine text based interfaces to make them efficient. guis for distribution software are still quite new. In time, with a lot of reworking, they will become more efficient. More importantly, users will be able to modify them in ways that will allow each company to fine tune form navigation to make it even more efficient.
• A few years hence, we can expect to see the first practical systems with speech recognition. Users will be able to speak names, part numbers, and commands the computer recognizes, making data entry twice as fast as it is today.
• Faster and more accurate forecasting algorithms with further reduce excess safety stocks and increase inventory turns, because modern computers can perform mathematical calculations fast enough to apply sophisticated probability theory to routine forecasting chores.
• Pricing will be more flexible because the user will be able to modify the actual code that calculates prices, instead of just the tables.
• Numerous sales and order management enhancements will “bubble up” from the user base as they try out ideas and share their ideas with other users.

The biggest breakthrough in the next generation of distribution software will be specialization. As systems become more integrated, the size of the application code shrinks and becomes easier to modify. Systems that were beta tested last year have about 90% less code than the systems they replace.

Because the relationship between code size and the effort required to create it or modify it is exponential, a 90% reduction in code size translates to a 30-fold reduction in effort. This means an enhancement that would take six weeks of work on a current generation system can be completed in a day on a next generation system. With modifications made this easy, users, vendors and third-party programmers and consultants will be able to customize systems to the needs of smaller niches and even individual users.

Of course, programming is only part of the equation -- it still takes time to think, plan, design, test and document changes.

Perhaps the most significant change in the next generation will be the reemergence of users in control of their own software. As the software becomes simpler, more open and easier to modify, we’ll see a return to users have control of their own destinies by developing their own, independent support resources. In the future, users will decide what their software will do.                   

Clark Yennie is the creator of Results, shims, Eclipse, the Roche Laboratory Manager and other software products. He is also the  author of Covestone distribution software and president and ceo of Covestone Corporation. Covestone has been under development for more than 10 years. It represents the cutting edge of software for distributors. Covestone will be released in early 2008. For more information, visit www.covestone.com or contact info@covestone.com.