IT103Microsoft Windows XP/OS Chap16

Editor's Notes

• #3: Too often, performance is considered last. As you progress through this chapter, emphasize that performance depends on proper planning. Each application that runs on a system requires certain resources to run efficiently. By analyzing a system’s workload and the requirements of each application, you can size systems appropriately. This chapter teaches students how to monitor system performance, locate bottlenecks, and take action to improve performance.
• #4: Perhaps the most important aspect of performance management is ensuring proper performance from the beginning. Too often, users purchase a new system with no clear idea of whether it will support the applications they plan to run. They might check the resource requirements on the software packaging and see that their system meets the minimum requirements. However, they might not consider how far from optimum those minimum requirements are. Describe how to sum up the requirements of the applications that will run, and factor in other elements such as planned upgrades and future applications.
• #5: Adding memory is often the least expensive way to improve a system’s performance. Multitasking with multiple applications often results in the use of virtual memory to satisfy the requirements of the system and the running applications. Virtual memory is much less efficient than hardware memory, due to the slower access time of hard disk drivers compared with RAM.
• #6: A shortage of memory often masquerades as inefficient disk access times. When a system is using virtual memory heavily, less disk time is available for common read/write tasks. Disk fragmentation also contributes to poor performance, and slow network links degrade system and application performance because applications have to wait for data files to load over the slow link.
• #7: In addition to speed, factors such as the amount of level 1 and level 2 cache and the design of the supporting chipset affect CPU performance. Point out the common misperception that CPUs need to be faster even though most systems regularly run at less than 10 percent CPU utilization.
• #8: This slide illustrates the additive effect of multiple applications on a system’s resource requirements. Encourage students to consider the uses that will be required of a system. Ask them what will happen if a user decides to have four instances of Microsoft Excel 2003 running at once.
• #9: This section explores the use of the Performance console to monitor performance. Offer real-life scenarios in which these tools might be used to analyze performance issues.
• #10: This slide depicts the default configuration of the Performance console. Use it to give students a high-level view of the console’s features. If possible, launch the Performance console on a classroom system to demonstrate the features.
• #11: System Monitor is the real-time monitoring component of the Performance console. By adding appropriate counters, you can watch a system in real time to see what is affecting performance. If you have opened the Performance console, include a tour of the histogram and report views.
• #12: This slide depicts the addition of available memory to the list of counters in System Monitor. Walk through this process in class, if possible. It is critical that students understand the configuration of counters and charts for effective performance monitoring.
• #13: Demonstrate the various performance objects and counters, and read the explanations included with them. Display the resulting chart, and use the System Monitor Properties dialog box to change chart colors and the sample rate.
• #14: This slide depicts saving and later using a System Monitor view. A saved view uses an ActiveX control (shown in the last frame) to access performance data from the system and displays it in a browser window.
• #15: Performance logs record performance data at regular intervals for long periods of time. The resulting logs can be viewed and reported on in System Monitor. We will demonstrate the creation of a log in the next slide.
• #16: This slide shows the creation of a performance log from a saved view. If possible, configure a performance log in class and start it. Allow some time for the log to record some data and then continue with the next slide.
• #17: You can use System Monitor to view a performance log. System Monitor can also present the data in the histogram and report views. By changing the time range, you can drill in on specific events in the log. View the log you captured earlier to demonstrate this feature.
• #18: Performance Alerts notify users or administrators when certain counters have reached or exceeded a value defined by the administrator. You can use them to log performance data during critical events and to identify issues that require attention.
• #19: This slide shows the creation of a performance alert. Alerts are a great way to start performance logs only when there are meaningful events to log. If time permits, create an alert in class to demonstrate this feature.
• #20: This slide steps through the Task Manager utility. Bring up Task Manager in class, and demonstrate features such as starting tasks and switching tasks. Show students how to sort the Processes tab to show the greatest users of memory or CPU resources. Use Task Manager or the frames of this slide to point out methods for ending tasks and processes, tracking network usage, logging users off the system, and sending messages to users.
• #21: Improving system performance is often a circular task. It requires monitoring performance, deciding which resources need improvement, and then analyzing the improvements. If the change does not achieve the expected performance benefits, you might have to repeat the monitoring phase to determine which additional resources need improvement.
• #22: These three counters give a reasonable picture of whether a system needs additional RAM. Discuss why a low Available Bytes and a high Pages / sec or a high Page File % Usage reading might indicate a need for more RAM. Also note that a high Pages / sec value might also explain poor disk performance.
• #23: These two counters can be used to monitor the performance of a physical disk. Explain that a high Average Disk Queue Length value indicates that processes are having to wait for disk access. If % Disk Time is also near 100 percent, the disk is not keeping up with demand.
• #24: This slide shows the use of the Scheduled Tasks Wizard to create a scheduled task to defragment a system. This important disk management task can provide noticeable improvements in disk performance. Ask students what other tasks might be scheduled to help maintain a system.
• #25: The options for directly improving CPU performance are limited. Most involve replacing the CPU, the motherboard, or both. You can add a second CPU to a dual-processor-capable system, but this might be almost as expensive as replacing the CPU with a higher-performing model. Discuss the myths and merits of hyper-threading CPUs. Be sure to point out that, while they appear as two CPUs to the system, they still use a single execution engine, but keep it busier by managing parallel application architectures. This achieves improved performance for well written multitasking applications and applications compiled for hyper-threaded processing.
• #26: Intel, AMD, and Transmeta all have mobile processors that adjust their speed according to power conditions and system load. When the CPU is in its slower modes, performance might decrease noticeably. You might be provided with a utility to control the processor’s speed, or you might have to change the system’s power scheme to Always On to force the CPU to remain in high-speed mode. Take into account that overriding this feature will affect battery life.
• #28: Insufficient memory, underperforming disks or network links, and underpowered CPUs all contribute to poor performance. By using the Performance console, you can determine which resources need improvement and upgrade them.
• #29: Performance analysis and improvement might require several iterations before meaningful performance gains are realized.