ARCHITECTURE TO BUILD ON 1995

For retailers and wholesalers, the migration to client-server platforms and open-systems environments may be the single most important step in the management information systems arena. The technology provides the strategic decision-making tools that are vital to competing in today's fast-paced, information-driven food industry.

electing a new generation of hardware and software systems and educating and training a wide range of end users at all levels of the organization.

From the biggest chain and wholesaler to the smallest operator, switching to an open-systems computing environment in which information is easily, accurately and quickly available is crucial. The issue is how to get there successfully and cost-effectively. Here is how SN's discussion with five leading MIS executives went on the subject:

SN: How important are client-server and open-systems computing today?

PATRICK STEELE: Changing computer architecture and the need for open-systems and client-server platforms are as necessary in the supermarket industry as in any business today. Our industry is faced with the same issues as anyone else: How can technology be leveraged to provide the necessary capabilities for each business to prosper?

Information systems have to be tied in with bottom-line objectives. That must be the main role of information systems in any organization today, and that is as true for the single-store operator as it is for the largest chain or wholesaler.

RAY HAMILTON: This whole area is of key importance today. Every newspaper is filled with articles about the amazing things going on with computer architecture and the desktop PC. It's driving what we are doing in the industry. Five years ago a 286 PC cost $5,000. Today, the price-performance ratio has changed so much that it's making everything a possibility.

The desktop PC is now the primary computer in the hands of retailers. It is going to see more penetration, and managing that environment of networked PCs in our stores and our corporate offices is a big challenge for us.

PETER ROLANDELLI: Computer architecture is a tough issue today because it involves so many factors: hardware, software, communications, platforms, whatever. This whole area is much more complex than it used to be. The controlled batch-processing world in which a mainframe is connected with a terminal to print out invoices and collect POS data was very simplistic compared with today.

Client server is a tool. Companies don't necessarily start out trying to convert to client server. They start out trying to solve a problem. The trick is that client server often winds up providing better and quicker answers in many cases. Nobody has patience anymore to wait a long time for answers, and rightly so. It's a very competitive industry, and we need to provide rapid solutions. Client server does that.

DAVID HAYES: Client server is throwing a level of complexity that information systems managers have never had before. Users are clamoring for data that they can interrogate on their own. In stores, we see all of the open architecture that's becoming available at a fairly low cost compared with some of the proprietary systems.

BILL MAY: The architectures are moving away from mainframe-centric to client-server systems. An important benefit is we don't have to have intervention from technical staff to write a program to extract data. We can now have nontechnical people using software programs to access data in their own way -- data that they need for making immediate business decisions.

The trend, though, is going to take a lot of time and money. I would suggest that the mainframe is going to be an important player for a long time to come. What the industry primarily is looking for in client-server is potential savings.

The catch is that, yes, a client-server type system may be less expensive to purchase. But there are some tremendous cost issues involved in training end users in the organization how to use the technology. There may suddenly be a large number of people with a huge amount of hardware and software computing power at their desks.

STEELE: The decision whether to move toward open-systems or client-server architecture must be made on a chain-by-chain basis. Client server satisfies the need for rapid applications development and systems capable of providing immediate information. The major companies are headed in that direction, and some are already well along the technology curve.

Smaller companies are also moving in this direction and in fact can sometimes make the transition even more easily because they may have less of an investment in existing information systems and technology.

ROLANDELLI: Open systems allows me to obtain more solutions for my problems, so why wouldn't I move in that direction. With open systems, a company is not tied to any one vendor's program. Open systems means more solutions.

SN: Has the migration to open-systems architecture proceeded as rapidly and smoothly as expected?

ROLANDELLI: I think solutions are starting to take shape much better and, at the same time, people are catching up in terms of their technical knowledge. A lot of architectures involve new concepts and were the focus of attention of just a few people. Now more and more people within organizations are learning about these technical areas. We have elevated ourselves to a new technical platform.

We have taken a step up and now may even flatten out for a while as we learn to utilize these tools more fully.

STEELE: I don't think the reality of open systems has been as pure as many people thought it would be five years ago, but it certainly has come a long way. We are now seeing POS systems at the front end using PC-based registers. The PC has been a major catalyst in this area.

And as new peripherals, whether they be scanners or printers or displays, work with standard PC connections, it makes it a lot easier for companies to move forward and invest in technology. With proprietary systems, companies became locked in one direction, and if they wanted to change it required very large capital investment.

HAMILTON: Overall, I feel comfortable with our success. I define open with three terms: portable, scalable and interoperable. Open means being able to mix and match product, upgrade hardware without changing software and being able to move software to a different computer system. That is the ideal. But once you start mixing and matching vendors, other challenges begin appearing.

ROLANDELLI: It is almost like the choices in a department store, where there are 100 different styles of shirts and you almost begin to think, "I wish everybody wore white so I wouldn't have to worry about which color to choose."

There are so many pieces to put together now, and it is our responsibility to pick out all the pieces. It is a lot tougher than it used to be. The positive side, though, is that if a solution doesn't work, you can change a piece of it. The whole system or program doesn't have to be replaced.

SN: What do you see as the major obstacles to moving to client-server and open-systems environments?

MAY: People have to be trained how to use the technology. Also, there has to be enough support staff to provide technical help when end users run into problems executing a program or when a software program or file server doesn't work right. The support staff has to be at hand to correct those problems.

People often look at the hard costs involved in purchasing the hardware and software but fail to take into full account the soft costs involved in training and supporting end users in a client-server environment. Will the support staff be capable of handling all the technical issues that arise in a client-server environment? That is a huge, huge issue.

HAMILTON: Traditionally, in the MIS department, we have managed centralized mainframe systems or midsized proprietary systems, with some connectivity. But today, with desktop PCs in our offices and in every checkout lane in the store, we are managing a very different environment. We have to take advantage of the tremendous value of PCs, we have to have them connected on a network and we have to be able to support them. That's a huge task for us.

The challenges are, how do we keep the PCs maintained physically, how do we keep software up-to-date and how do we get the right software onto the desktop? Probably the biggest challenge in a client-server environment is, how do we get the software to the workstation effectively and cheaply?

With the speed technology is evolving today, it is difficult for a large chain to keep the right software on the desktop for serving a particular business environment. Having the latest PC may not be necessary, but having the appropriate software is.

ROLANDELLI: We now scatter information processing to servers, to in-store processors and, in some cases, to outside processors. Controlling that world has put a large strain on our data processing organizations. I don't think we as an industry have truly come to grips yet with just how large of a support component is involved.

The need to support these systems has eaten away at the time available for developing new ideas. Clearly, with new concepts comes the need for support, because you don't just get rid of the old system right away. Typically, there are both old and new systems to support. There are also many more components to support today. There may be eight or nine communications media, a wide area network, new servers and data bases, all of which require support.

STEELE: People have to view it, like many things, as a journey, and a long journey. It can't happen overnight, both from a capital investment and technology standpoint. It is not fully there yet, but it is moving in the right direction.

MAY: I would tell anyone not to rush into client server before carefully weighing the cost, the staff retraining and the support tools needed to run the hardware and software systems. Another key is that you just don't go from a mainframe computer today to client server tomorrow. There still has to be a linkage maintained to the legacy mainframe system, which is where most of the critical data in almost any organization today resides.

I would suggest that many companies, although they are making a transition to client-server computing, will probably never totally leave the mainframe environment. If nothing else, the mainframe will be turned into a huge data repository or file server of tremendous size.

One example of that is electronic marketing that relies on daily scan data generated from retail POS systems. That is a tremendous amount of data, which could most efficiently be stored on a mainframe computer serving as a file server rather than trying to contain it on some smaller platform. So I believe the mainframe computer in many organizations will still have life.

SN: Are there particular areas where open systems are especially important?

STEELE: That depends on each company and its particular need. If one company is doing a lot of work with back-end systems and feels open systems will enhance or satisfy those needs, then that is where the most important emphasis will be. If another company is working on in-store applications and wants to migrate to an open platform for POS processing, then that is the arena that will be most important. It depends on where each company is and what the priorities are. But open systems are needed and important in all areas of the business.

MAY: Open systems are especially important for decision-support systems. We now have software packages on desktop file servers that provide employees with access to a large amount of information. Employees can write queries and obtain data from virtually anywhere within the organization, from any allowable file server. People now have the power at their fingertips to use executive information systems or decision-support systems -- and that is an area where client-server and open systems are vital.

But, quite frankly, this is so new to our industry, it is just being defined. There are still a lot of unanswered questions. We are just discovering ways to use the desktop power of client server for competitive advantage.

SN: Could you address the issue of computer architecture from the perspective of a wholesaler working with independents?

MAY: At Spartan we serve roughly 500 independent stores and 325 owners. We see every type of system and every configuration you can imagine. Our challenge is to set up an open-systems environment here and persuade our independent operators to move toward open systems so that we can connect to them and move data up and down the pipeline in a timely and cost-effective manner.

The difficulty in that is that many retailers buy proprietary systems that are difficult for us to connect with. Our challenge is to help them understand that the easier they make it for us to connect with them, by moving to open systems, the better off they will be from a competitive point of view. They will be able to move marketing, customer demographic and product movement data back to us much more efficiently and accurately. We, in turn, can turn that data around and provide decision-support systems to help them make decisions.