Introduction

The astounding growth of computer networking in recent years has been characterized by the struggle of network managers to keep pace with the demand for network resources. As more and more users, running more and more applications, demand increased network bandwidth, network managers continually find themselves adding more expensive capacity in an effort to satisfy the needs of their customers. Even still, networks remain congested, and applications continue to run slowly. New breeds of aggressive applications that use networks compound this struggle.

In general terms, performance of networked applications depends on complex interactions among applications, servers, and networks. IT organizations need a detailed, quantitative understanding of these interactions to efficiently and cost-effectively troubleshoot and deploy applications. Determining whether business applications are functioning properly from the end-user perspective is the single most important challenge faced by IT in delivering acceptable service for business-critical applications.

Application performance management (APM) is both a business concept and an evolving technology. Conceptually, APM is based on the recognition that network users are concerned only with end-to-end services. In today's business world, it is not practical to take the view that one person can control or manage an entire end-to-end scenario. The key to APM is to present, view, and share performance data from the user, through the network, and to the server in a language and view "common" to all parties.

In an e-commerce system example, if a user clicks a link to place an order, and the system does not respond within 10 or 15 seconds, the user is likely to go elsewhere. Exactly where the system performance shortfall is (such as the end user's ISP, a misbehaving router, or insufficient bandwidth) is of no concern to customers. APM begins with organizations defining which of their business processes are critical. For example, an organization might say, "Our web page must be available to customers." They would then define service objectives such as, "The web page must be available and load within 15 seconds, 98 percent of the time on weekdays and 90 percent of the time on weekends." If there are no clear-cut service objectives in the first place, management is meaningless, and performance evaluation is irrelevant.

After an enterprise has decided to look at the network from the outside, in terms of business objectives, the IT department is faced with linking those objectives to the technical realties.

Unfortunately, the traditional network management tools available to IT (whether they are aimed at troubleshooting, configuration, fault management, or even security) tend to be reactive rather than proactive. By the time they provide an alarm, it is too late from the point of view of the disrupted business process.

This book takes a practical look at how APM can assist in the overall management process by proactively identifying hotspots to various groups (such as server, desktop, and network) within an organization. By completing an overall view specific to a business transaction, readers will be in a position to define and meet business requirements relevant to their applications, which will subsequently optimize their overall delivery processes.