Printer Friendly

Should you migrate to client/server computing?

IBM is the latest vendor to embrace client/server computing by creating a 900-staff member unit that will work closely with users and other newly created IBM business units to develop software solutions for the new computing environment. But does IBM's endorsement mean that all organizations should start migrating to client/server computing?

Conventional wisdom states that larger firms certainly should be moving to an open, network client/server architecture to handle their information processing needs. But does this apply to smaller companies as well, and what pitfalls await the unwary in organizations of all sizes?

No one disputes that client/server computing is complex. Proponents argue, however, that the benefits far outweigh the difficulties.

Ideally, client/server computing gives you the responsiveness, familiarity and ease-of-use benefits of PCs with the performance, reliability, data integrity and security of the mainframe, coupled with the ability to support many users simultaneously. Since the processing is distributed, client/server applications benefit from the increased memory and processing power of the server.

Client/server computing evolved from the limitations of early LANs which used PCs as shared, centralized file servers. PCs were designed to support single users and were illequipped to handle the throughput levels, and information sharing and management tasks required of a server. File sharing between the server and local PC was slow and cumbersome, and data integrity and security limited. For data base applications, the limitations were even more severe.

Specialized network servers, optimized for intensive input/output multiprocessing and data integrity, have overcome the limitations of PCs as servers. The next step towards client/server computing is to divide the application, putting part of it on the user's computer, the client, and the rest on the server.

While this division is needed to leverage the power and responsiveness of the client computer, it creates a problem when implementing client/server computing at an enterprise level. That's because no one has defined a standard or consistent approach to deciding which parts of an application should go on the client and which on the server.

Suppliers of client/server data base management systems typically do most of the processing on the server, limiting the client to some preliminary processing. Many client/server development tools build LAN applications on top of these data bases and follow the same processing split. However, a significant and growing number of such tools support networks where clients also function as servers. In these cases, the processing tends to split evenly across all the machines on the network.

Adding to the complexity, client/server network operating systems usually perform most of the applications processing on the client computer, using the server for storage and to perform security and operating system functions. Further, each client/server application developed in-house may have its own processing split.

It's hardly surprising, then, that most client/server networks have been implemented on a local rather than enterprise level. Even so, the migration is picking up speed. Forrester Research Inc., a market research firm in Cambridge, Mass., estimates the market for client/server hardware and software will grow from $4.9 billion in 1992 to $38.3 billion by 1995.

Initial reaction to IBM's creation of a client/server computing group has been guardedly optimistic. Tapping the client/server market is considered critical for IBM as the mainframe continues its decline, but the road ahead will be tough for the company. Most of the tools for creating distributed applications have been sold by the relational DBMS vendors, such as Oracle and Sybase, who have an edge because of their ability to optimize the development software for use with their own data base systems.

Typically, these vendors store some of the application code on the server to increase performance, since the server is usually more powerful than the client. Also, the server software can be modified as needed without having to alter the client-based application code.

Consultant Shaku Atre of Rye, N.Y., advises buying the most powerful server your budget allows since it will be doing the bulk of the data base work. As a rule, data base processing should occur where the data resides. It makes sense to put personal, single-user processes on the client, and shared, multi-user ones on the server. Also, to minimize network loading, clients should be powerful too and should validate data themselves and only go to servers when necessary.

Next month, we'll examine how to choose the right network server.

[Data communications consultant Morris Edwards serves as program chairman of the Network Computing Solutions Conference and Exposition, or NetCom, which will be held at the Radisson Centre in Miami, FL from March 30 to April 1, 1993, and the Disneyland Hotel in Anaheim, CA from May 4-6, 1993.]
COPYRIGHT 1993 Nelson Publishing
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 1993 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Title Annotation:NetComm Update
Author:Edwards, Morris
Publication:Communications News
Article Type:Column
Date:Mar 1, 1993
Previous Article:TCA elects board, chooses officers, honors members.
Next Article:Just how application-savvy are you?

Related Articles
Client/server religion: should you convert?
World leading client/server banking system available as a service bureau solution for community banks in Indiana and Michigan.
Thin Client Computing Saves Time and Money.
Global namespace: The Future of File System Management, Part 2. (Storage Networking).
Enterprises shifting to the web: the case for SSL-VPN.
CaminoSoft announces Information Lifecycle Management solution for Windows XP workstations.
MySQL explained.
Collapse of the 'Web tier': application front-end products condense functionality into a single, more elegantly designed platform.
Controlling your migration to IPv6: a gateway to tomorrow.

Terms of use | Copyright © 2016 Farlex, Inc. | Feedback | For webmasters