Network operation is a critical component of any Disaster Recovery and Business Continuity Plan. Historical data shows that failures are caused by serveral factors.
Network operation is a critical component of any Disaster Recovery and Business Continuity Plan. Historical data shows that failures are caused by serveral factors.
More than ever, software applications enable the language of commerce; companies of every size, in every industry, depend on enterprise applications to execute virtually every aspect of their business in today’s global marketplace. With the average Global 2000 corporation 
operating between 250 and 500 packaged and custom applications
1, genuine risk lurks beneath the façade of a well-oiled software machine.
Companies spend millions of dollars implementing enterprise software, but after their deployment, many applications are minimally managed until they are stricken by downtime. In fact, application problems are the single largest source of IT downtime
2; the analyst firm Gartner estimates that 40 percent of unplanned downtime is caused by application issues. The business impact can be devastating – among Global 2000 companies, application downtime costs each organization an average of $40.7 million per year, or 3.6 percent of revenues.
Network performance management solutions typically measure and report on the four factors that most directly affect application performance in a network environment:
- Bandwidth Consumption measures the bandwidth consumed by each application task. If the sum consumption exceeds available bandwidth on a given link, adjustments need to be made to bandwidth or non-essential traffic. Most application performance management tools provide these measurements by five- or fifteen-minute intervals. However, application tasks cannot be equated to these intervals since dozens of tasks could occur during an interval period. This data is nonetheless extremely useful in gauging, for example, whether an application is bandwidth-sensitive by dividing the total bandwidth for an interval by the number of application turns.
-
Application Turns are an extremely important application metric. Each request/response pair on a network is called a “turn.” For each turn, the application must wait the full round-trip delay of the network between the client and the server. The greater the number of turns, the slower the application will perform. Excessive turns on overburdened networks further slow down the network and are extremely detrimental to application performance. Therefore, lowering the number of application turns can dramatically improve network performance and, subsequently, end-user application response times. -
Application Sensitivity uses bandwidth and turns information to help network managers determine whether the application’s sensitivity is to bandwidth or latency factors. A bandwidth-sensitive application passes large amounts of data between client and server, and can be identified as such if increases or decreases in bandwidth significantly change the application response. Latency-sensitive applications can be measured by the number of round-trip turns (RTT) required to complete an application task.
-
Application Efficiency is determined primarily by protocol efficiency factors including a small TCP window and small frame sizes. Other elements that impact an application’s effect on the network include low utilization of memory caches on client machines, and unnecessary data transfers.