Enterprise computing has long been associated with a big mainframe in a backroom, pumping through vast amounts of data and thousands of transactions a day in a non-stop, fault-tolerant way. In many companies, that's still the case. Mainframes continue to sell. They're becoming less expensive and Internet compliant.
"Mainframe MIPS increased by 55% last year," says Mitch Gross, president of Mobius Management Systems (New Rochelle, NY 914-637-7200. "Gartner Group will tell you for every four applications that are going from System 390-based servers to Unix or NT, eight are going in the opposite direction and at an ever-increasing pace. The big applications are all going to big machines. We did an application for a customer who had 6,000 CSRs accessing images continuously. You need a big System 390 to do this. The communications backplane of a small server isn't big enough to support high volumes."
Bob Schwartz, marketing manager for IBM's (White Plains, NY 914-431-7676) EDMSuite, agrees: "The big powerful box makes it simpler," he says. "You're managing one box, although you can have multiple object servers. When it comes to scalability and manageability, mainframes do it better. People are discovering that the mainframe is pretty competitive."
On the other hand, Microsoft's NT server is selling like hotcakes. It's making inroads into the big enterprise applications.
"We have our sights set on moving into mission critical applications where you need robustness and scalability," says Richard Noffsinger, document management / imaging / workflow industry manager at Microsoft. He says the next version of NT will be much more robust and fault-tolerant than 4.0, thanks to Microsoft's agreement with Tandem. "Today NT scales well to four processors," Noffsinger says. "The next version should scale well to 12 processors." Microsoft has also developed clustering software that distributes work among several servers.
Imaging vendors are writing more enterprise imaging software for NT. "NT itself has a lot of grounds to go forward, although it has definitely got to prove itself," says Walter Aallen, support manager at Siemens Nixdorf.
"We're trying to get away from the big box mentality," he says. "The large box will always eventually run out of cycles. No matter how big you grow something it will always grow bigger. It becomes a performance issue based on document retrieval time. If four to six seconds is acceptable to retrieve a document you might be able to run all your applications off one machine.
"If you're going to be working with a high-volume installation and response time isn't one of your major concerns, you don't have to build the system too big. We want you to add more technology to your system's backbone as your system grows instead of adding more power to a single large box.
"You can put more devices on the system to address your requirements or bottlenecks as you need them. If you're adding more clients, you may need more client servers. If you need more optical storage, you may need more optical servers. If you need higher document throughput at your scanners, you may need more scan modules."
