Balanced Computing and the Server’s Changing Role

In balanced computing, processing is shared between clients, servers, and any other devices connected via a network. This distribution is inherent in the design since applications are already spread out on different machines and connected through web services. In balancing, the processing required by each new use is often shifted back to the user’s system, thereby taking fuller advantage of client-side processing power. This allows for improved scalability since the processing load is increased insignificantly by the addition of users.

Load balancing can also be achieved by building service-oriented applications (SOAs) where components run on different nodes in multiple locations duplicate services on multiple nodes. This duplicating of services on multiple nodes also improves reliability since there is no single point of failure that will topple the entire application. Through balanced computing, platforms can take maximum advantage of computing and display capabilities on each platform. The virtual application which is balanced across multiple nodes remains transparent (its complex functions hidden) while the user utilizes his or her own collection of devices to run and view the application on the user end.

Balanced computing not only distributes the processing load, but changes the role of the server as well. Instead of computing so heavily, the server primarily directs traffic. Given rich clients and decent Internet connectivity, users directly contact databases instead of requiring server intervention. Rich clients are applications in user computers that retrieve data through the Internet. As previously discussed, the proliferation of web-based applications replaced the user interface with the web browser. Scripting languages, like JavaScript or VBScript, were embedded into HTML to improve user interfaces (among other things). Java applets were also added. But nothing could compete with the user experience of using an application built from its local environment. Developing technologies like improved web-page scripting languages and AJAX (Asynchronous JavaScript and XML) made web browsers function more like rich client applications.

Another method used to reduce demands on the server uses a connecting device to re-direct previously server-side processing to the client. This depends on the client device capability and Internet connectivity. If these are weak, the server picks up the slack, making application performance consistent on both the client and server ends.

User Experience and Development

Balanced distributed computing models improve user experience and expand development. Developers can focus on user experience by examining devices and customizing features. For example, different user interfaces can be customized for different departments that require different resources to perform their function. Different roles would have their own user interface. Well-defined user roles and profiles that are stored on user machines make more efficient use of server computation. It allows the server to pull customized responses based on the identity/role of the user. To further reduce server demand, clients can communicate directly with databases by requesting information for the user role profile. This eliminates the web server as middleman for the request and computation on the output side.

Data integration on user platforms offers new opportunities to build applications that draw data from a variety of sources and can add different contexts. In a balanced distributed computing model, web services send information that is usually stored on databases or servers (like financial information) to the user’s machine. It accomplishes this by using the client-side’s processing power. These responses are formatted in the increasingly popular, universal XML. Desktop applications (on the user’s system) can take that information and analyze it in different contexts.

The decentralization of distributed browser-server models also improves security and protects privacy. For example, data repositories are often located in a different location from the server. This makes it more difficult for external attackers to find. It also makes it less accessible to internal attackers. Also, it is safer for user profiles to be stored on individual machines, rather than on a central database.

Distributed computing models address the future of IT architecture and application. Organization must aim to create independent and flexible applications that can respond quickly to a variety of contexts. Connections must be agile. Loosely coupled applications, characteristic of distributed computing models, withstand broken connections and slow Internet performance. This protects core technologies from customer demands and lack of Internet bandwidth.

TrackBack URI for this entry

Comments (0)

Subscribe to this comment's feed

Show/Hide comments

Show/Hide comment form