A Strategy for Managing Performance
Computer systems play an important role in business, and with the appropriate attention paid to the compute environment, compute costs are minimized and compute equipment utilization is maximized. The result is a direct benefit to the bottom line. An important and often neglected area of compute environment management revolves around the tasks of managing computer system performance. It's easy to understand why this task is neglected, because poor performance and identifying its causes can be ambiguous and time consuming. What's needed is a strategy for managing performance.
This article offers suggestions for developing a strategy for managing performance. The strategy is high level so it can be applied to almost any compute environment. In addition, a variety of performance related products are listed.
This article is intended for CIOs, CTOs, IT managers, service delivery managers, and anyone interested in managing performance of computer systems.
Performance Versus Availability
The goal of the compute environment is to deliver adequate and reliable services to support the company and customers. Anything that adversely affects availability of the computer systems is immediately felt and quick remedies are sought. Various figures exist for the amount of lost revenue from an hour of downtime of critical computer systems in different industries. For example, a financial brokerage house can lose as much as $6.5 million for each hour of downtime.1 To avoid this agony, extraordinary amounts of money are spent for equipment and personnel to avoid downtime and increase availability.
But what about the less obvious performance decline? Poor performance of computer systems has a negative impact on company revenues along the same lines as downtime. As with downtime, there is a negative effect either through lost opportunities (customers or orders turned away), or through lost revenue (inability to ship or bill). A comparative cost figure can be derived for performance problems; a five-percent decrease in performance of a critical application translates to just over an hour of lost processing in a 24-hour period. For the same brokerage house, a five percent performance decline can also result in millions of lost dollars. If the loss were due to downtime, a remedy would be immediately implemented, but for poor performance, without a strategy to detect and correct the decline, the impact on revenue might go unchecked indefinitely.