Archive for October, 2009:
Using Performance Logs and Alerts (9)
Alerts work by monitoring performance counters on your system, so you need to add at least one counter to the Counters list before you can do anything else. Click Add and you’ll see an Add Counters dialog box. I chose processor idle time as an example because it’s easy to generate an Alert using this counter. Figure 7.12 shows the General tab for the sample Alert. An Alert can rely on more than one counter, but you should create one Alert for each major event so that you can detect which Alert went off.
You also need to select an Alert condition. For example, if you’re monitoring memory, you might want to set an Alert when the amount of available memory falls below a certain level. As you can see, this tab also allows you to set the monitoring frequency and the Run As field.
The Action tab contains options for telling you when an event occurred. The following list discusses each option and explains what they mean.
Log the Event to the Application Event Log This is the best solution for noncritical alerts. You want to know the event happened, but you don’t have to fix the condition immediately.
Send someone a message Reserve this option for critical alerts that require immediate attention. For example, if a lack of memory or hard drive space will cause the system to fail, then you want to send someone a message. There are two odd requirements when sending a message. First, the message must go to a machine other than the current machine. Second, the message must go to the machine and not the person. If more than one person uses a machine, this second requirement could become a problem. If the message sending option looks like it won’t work, then you could always write a script to send the message in another way.
Taken From : Microsoft Windows XP Power Optimization
Using Performance Logs and Alerts (8)
One column contains the Thread IDentifier (TID) for the process. Windows XP assigns every instance of an application a unique TID. For example, if you open two copies of Notepad, they’ll both have a different TID. It’s safe to ignore this column for troubleshooting and performance needs.
You’ll also see columns for clock time (indicates the process priority), user time, and kernel time. These are the three columns of most interest for performance needs. The priority tells you how the application shared CPU cycles with other applications. If you see an application that sets a high priority, you might want to explore further and ensure the application warrants the extra processing cycles. The user time indicates how much time the application spent in user mode, generally serving user needs. The kernel time shows how much time the application spent in protected mode accessing system resources.
Creating Alerts
You’ll use the Alert folder to warn you of system conditions or to tell you when an optimization event has occurred. For example, you can use it to warn you that system memory or hard drive space is low. The alert could occur because an optimization you performed left the system too short on a specific resource. Disabling the paging file makes your system faster and reduces hard drive resource use, but at the cost of memory. When you use the technique located in the “Disabling the Paging File” section of Chapter 6 to disable the paging file, you might also want to set an alert to tell you whether the cost of the optimization is too high. Here are the techniques that an alert can use to warn you of impending disaster.
? Create event log entries
? Send you a console message
? Run an application as part of the alerting process
To create an Alert, right-click Alerts, and then choose New Alert Settings From or New Settings Alert from the context menu. Type the name for your Alert in the New Alert Settings dialog box and click OK. You’ll see a properties dialog box.
Taken From : Microsoft Windows XP Power Optimization
Using Performance Logs and Alerts (7)
Viewing the Trace Log
After you collect the required data, stop the Trace Log from running. Right-click the Trace Log entry, and then select Stop from the context menu. The indicator will turn red. Open a command prompt in the \PerfLogs directory. Type TraceRpt
NOTE TraceRpt accepts other command line switches, but you normally won’t need to use them unless you perform multiple traces on one system and don’t want to overwrite the default files. Use TraceRpt /? to see a list of other command line arguments and switches.
You can open the DUMPFILE.CSV file in a number of ways, but the two best ways are with a spreadsheet or database manager. This file will contain a lot of information and working with it any other way will prove difficult. Sorting the data is going to be a necessity. Figure 7.11 shows a typical DUMPFILE.CSV file output.
This log shows what happens when you select the default monitoring options. As you can see from the figure, I created a few new processes, but the system generated a lot more. Much of the trace data remains hidden in the figure. For example, you can’t see the event name or type columns.
Taken From : Microsoft Windows XP Power Optimization
Using Performance Logs and Alerts (6)
The Log Files and Schedule entries work just like those for the Counter Log. One difference that you’ll note is that Trace Logs use different log file types. You can’t read either log file type directly; both require the use of the TraceRpt utility.
The Advanced tab contains the buffer settings for the Trace Log. A Trace Log gathers a large amount of data. The log service saves data to temporary buffers in order to improve performance. If the settings you choose don’t meet minimum requirements, the log service will override them. Normally, the log service saves data in the buffers to disk when the buffers are full. However, you can also set the number of seconds between data transfers to reduce the risk of losing data.
TIP The settings you use for the buffers will affect the performance of your machine and the activities of the Trace Log. In some cases, providing a larger number of buffers or using a larger buffer size will allow the log service to work more efficiently and reduce system load. However, adding too many buffers increases the risk of data loss or corruption should the machine freeze or act in other unexpected ways.
Click OK. Performance Logs and Alerts will attempt to start the Trace Log if you used the default settings. Otherwise, you’ll need to start the logging process manually, or wait for the conditions you set to occur. If the Trace Log lacks the proper permissions to start, the indicator will remain red and you’ll see a warning message in the Applications folder of the Event Viewer indicating the problem. Fix any problems and restart the Trace Log manually.
Taken From : Microsoft Windows XP Power Optimization
Using Performance Logs and Alerts (5)
You can choose the types of events that a Trace Log monitors. A Trace Log can help you see the flow of data on a system or determine when an application is creating too many threads. By knowing that an application is creating too many threads, you can begin optimizing it for better performance and behavior (too many threads can cause crashes) by reducing the number of simultaneous tasks the application performs.
This part of the Performance Logs and Alerts snap-in creates special logs that you have to convert into human readable format using the TraceRpt utility. You can read the resulting output in System Monitor or within an application such as Excel (the preferred method in this case).
Saving the Trace Log
The process used to create a Trace Log is similar to that of a Counter Log. You begin by right-clicking Trace Log and choosing New Log Settings From or New Log Settings from the context menu. I’ll assume you’ve selected New Log Settings in order to create the Trace Log object from scratch.
When you see the New Log Settings dialog box, type a name for your Trace Log and click OK. You’ll see the properties dialog box shown in Figure 7.10. Note that I’ve selected Events Logged by System Provider to show those options clearly in the figure. You can also choose nonsystem providers to track. In most cases, you’ll need to provide a Run As entry to ensure the Trace Log works as anticipated because few users have the rights required to perform this low-level work.
Taken From : Microsoft Windows XP Power Optimization
Using Performance Logs and Alerts (4)
To view the content of a log file, open System Monitor and click View Log Data (the icon that looks like a database symbol). You’ll see the Source tab of the System Monitor Properties dialog box. Click Log Files (or database, if you saved the data in that form). Click Add, and then select the log file(s) you want to view. Note that you can add more than one log file to the list to see a longer interval. Make certain that all the logs come from the same Counter Log. If you think you might want to limit the interval viewed on screen, click Time Range.
Click OK and you’ll see the static data collected in the log. Of course, this means you won’t see the Time Bar moving across the screen—System Monitor only uses the Time Bar when it has active data to read. All of the other options that you use for active data apply when working with a log. For example, you can display a graph, histogram, or report version of your log.
NOTE One change that you’ll notice is that the content of the Add Counters dialog box will change. You can only view data that you’ve recorded, which means the list of objects and counters will change. In addition, you can only select counters for computers that you logged. Any other computers will fail to appear on the list.
Creating and Converting Trace Logs
Trace Logs perform detailed system monitoring. It’s almost never necessary to use this level of monitoring to optimize your system. In fact, you normally use this level of monitoring to fix major system problems—the type that give network administrators nightmares.
Taken From : Microsoft Windows XP Power Optimization
Using Performance Logs and Alerts (3)
The Log Files tab of the properties dialog box contains fields that adjust the output log. You can choose from several output formats including text, binary, and SQL Server. The last entry is helpful if you want to store the results in a database for complex analysis and archiving. The Example field shows what the output filename will look like; you can monitor this entry as needed. Finally, you can add a comment to the file and choose to overwrite the old log file each time you restart the
Counter Log.
You’ll use the Schedule tab to define a starting and ending time for the Counter Log. The Start Log field contains entries that start the logging manually or on a specific date and time. The current time is the default setting. The Stop Log field contains entries that will stop the logging manually, after a specific time interval (seconds, minutes, hours, or days), a specific date and time, or when the log is full. The default setting stops the logging manually. Finally, you can set the Counter Log to perform a task automatically when the log file closes. This parameter includes creating a new log and running an application.
After you set up everything, click OK. If you’re using the default Start Log settings, the Performance Logs and Alerts snap-in will begin the logging immediately. Otherwise, you’ll need to start the logging manually by right-clicking the Counter Log object and selecting Start, or waiting until the predefined conditions occur. Active counter log entries have green icons, while stopped entries have red icons.
Viewing Counter Logs
You have a log that you created of processor activity or some other performance object. Having the log and doing something with it are two different things. You need to know how to view the log after you create it.
Taken From : Microsoft Windows XP Power Optimization
Using Performance Logs and Alerts (2)
The counters and objects used with the Counter Logs are the same as those used for System Monitor. In fact, when you click Add Counters, you’ll see the same Add Counters dialog box, as shown in Figure 7.6. Click Add Objects and you’ll see a dialog box that only allows you to select a machine and a performance object. If you select a counter, then you’re selecting just that counter or even a counter instance. When you select an object, it includes all the counters within that object.
TIP You need to consider the trade-offs between counters and objects. An object will require more hard drive space to store and will consume more processing cycles. However, you’ll have all of the counters for that object, so you don’t need to worry about finding a particular piece of
data. Using counters is more efficient, but lacks flexibility. If you don’t think to store a critical piece of data, it won’t appear in the log later for analysis. Generally, it’s best to use objects when you can spare the hard drive space and lost processing cycles.
After you select some objects or counters, you can set the sampling interval. Short intervals produce more data, so you’ll get results that are more precise. However, using short intervals also makes the impact of performance monitoring more severe. Again, it’s a matter of balancing system resources and performance against the performance monitoring requirements.
Notice the Run As field on the dialog box. In some cases, the person running the performance monitoring software won’t have the credentials required to perform all tasks. The monitoring will fail, in this situation, preventing you from obtaining good readings. To overcome this problem, type the name of a person with the correct rights, click Set Password to display the Set Password dialog box, type the password twice, then click OK. The counter can run now because it isn’t relying on the user’s
credentials.
Taken From : Microsoft Windows XP Power Optimization
When she came
When she came to see me the next day, she remained ambivalent, even negative, about taking sertraline (Zoloft).
“I have to fight my problems without a crutch. I don’t want to become dependent on this medication,” she said.
“I never cease to be amazed by the number of people who religiously take their medications for heart disease, blood pressure, or diabetes, usually for years without end, and yet refuse to take a medication to treat a disorder that is affecting their most vital organ, the brain. I decided to bring up this point as a tactic to get Joan to take her prescribed antidepressant medication.
“Please tell me something, Joan. You take Prilosec for your stomach problems. And you’ve been taking it every day for how many years? Five? Okay, five years. Now would you consider that a crutch? As being dependent on the medication?’’
“No, well, that’s different,” she replied.
“How so?”
She paused, unable to give me a rational answer.
Taken From: The Memory Program How to Prevent Memory Loss
and Enhance Memory Power
Using System Monitor (5)
The Data tab shown in Figure 7.8 helps you configure the data display. You begin by selecting one of the entries in the Counters list. This view allows you to change the color of the data, along with the width and style of the line. The scale determines how the lines scale in comparison with the rest of the lines on screen. Using a larger scale can often bring out details about the data, especially if variations are small.
The Graph tab contains settings that change graph display elements. The display elements include the use of a vertical and horizontal grid. You can also give your graph a title and assign a name to the vertical element (the horizontal element is always time). The Vertical Scale properties offer another chance to optimize the graph display. You can set the graph so it displays the data a little larger or smaller than the default settings allow.
Finally, the Appearance tab contains settings that change the colors used to display certain elements such as the graph background and time bar. This tab also enables you to choose a new font for the display. The use of a different font may make the information more legible or easier to print.
Taken From : Microsoft Windows XP Power Optimization