【正文】 econds to interact with their corporate data. With their own puter on their desktop they do not need to wait for the mainframe to respond. All of this cheap puter power has also made it possible to support putation intensive graphical interfaces, which are much easier for users to understand. In order for Client/Server to work, the application program which used to reside entirely on the mainframe has been split into two pieces, the client piece and the server piece. The portion of the application that resides in the Client personal puter includes logic for the presentation of information to the user and mechanisms for accepting user input. It also includes logic for data interaction at the Client, such as changing the appearance of the graphical interface as the user make various choices. The Client software often includes corporate business rules that allow data to be validated before it is sent to the server. The portion of the application that remains on the Server is usually just the portion that stores data at a central location, accessible to other users. The once mighty mainframes have been reduced to database servers, while the rest of the application has migrated out to the clients. This migration has caused many problems. The Client machines have bee fat. As the plexity of applications has risen it has bee necessary to supply users with more and more powerful puters with faster processors, more disk storage, and more RAM. It is true that personal puter manufacturers have been able to deliver better and better puters for the same amount of money, but the constant upgrades required are costly for corporations. It takes more people and time to upgrade hundreds or thousands of personal puters than it used to take to just upgrade the mainframes. Setting aside the cost of Client equipment, the proliferation of corporate applications, including business logic, to hundreds or thousands of machines has bee a major cost factor for corporations. Where IT staffs used to be able to maintain all the corporate software assets in one central location, they now need to maintain corporate software spread all across the corporation, often housed in puters that are out of IT control. There are many estimates available that the cost of Client/Server puting is eight or more times the cost of the equipment alone. Several panies offer large, expensive systems that have the sole purpose of maintaining corporate software spread out all over the corporation. The issue here is not personalproductivity software such as word processors and spreadsheets. Personal software products have enabled knowledge workers to attain new heights of productivity. The issue is the maintenance of corporate software, such as orderentry and accounting. Imagine the difficulty of making a schema change in a corporate database, and synchronizing that change with the updating of thousands of copies of the corresponding Client software. It is true that the use of departmentlevel LANs can bring down the number of copies that need to be updated to hundreds, instead of thousands. The size of the LANs must be limited, because the large size of the fat client software requires a responsive LAN or else the users will just copy the enterprise software to their own machines, bringing us back to thousands of copies to be maintained. The task of making simultaneous changes to even hundreds of copies of client software is still a daunting one. The splitup of the corporate applications is different for Browser/Server . The user interface and the data interaction ponents are still run on the user machine, but the business logic usually remains on the Server, usually in a special Server called an Application Server. The user interface and data interaction ponents do run on the Client, but they are not ultimately stored there. They are most probably Java applets stored in a corpora te Intra Web Server, and they are automatically loaded into the Browser software on the user machine as they are needed. If a newer version bees available, then that version is automatically loaded. We have achieved the same benefits as with Client/Server, but without the enormous distribution and maintenance problem. The only software that the user machine needs, besides an operating system and personal productivity software, is a Web Browser that runs Java applets. Since much of the fat of the application resides in the business logic Application Server, the need to constantly upgrade the user machines is greatly reduced. There is no need to spend precious time and hard disk space installing corporate software on thousands of user machines, or hundreds of departmental LANs. Corporate software assets now all reside in corporate servers, under IT control. Updates to systems are now easily coordinated. So exactly what is Browser/Server technology? Browser/Server is an important adaptation of modern workcentric puting that optimizes the performance of mission critical enterprise applications deployed via the corporate intra and/or World Wide Web. Browser/Server acplishes these performance gains with customized multitiered infrastructure and application designs that produce production solutions with reduced work requests and increased user interaction. （二） SOFTWARE QUALITY ASSURANCE The activity of software quality assurance is closely related to verification and validation activities carried out at each stage of the software life cycle[1]． Indeed， in many anizations there is no distinction made between these activities． However，quality assurance and other verification and validation activities are actually quite separate， with quality assurance being a management function and verification and validation being part of the process of software development[2]． An