Overview
It may be necessary to capture and review performance data from a server hosting Archive Shuttle or Archive Shuttle Modules.  This article outlines the general steps to record performance data on Windows 2012.
Details
In general to get a feel for how busy a particular server it is necessary to collect the following data items over a period of several hours or even several days:
– Memory
– Process
– Processor
– Physical Disk
– Network
For each item it is necessary to capture all of the data points beneath it. This will allow maximum flexibility later when it comes to reviewing the data. Whilst this data can be easily added to an on-screen view, that doesn’t allow for future review or analysis of the information. For that purpose it is necessary to save the data to file. In order to do that perform the following steps:-
Launch Performance Monitor
Performance Monitor can be launched from the start menu, or by going to start -> run, and typing perfmon.
Set up a new User Defined Data Set
Navigate to Data Collector Sets and then User Defined. Right click on User Defined and choose ‘New’ -> ‘Data Collector Set’
Give the Data Collector Set a name which will make it easily identifiable in the future (because the collector set can be re-used)
Choose the option to create a data collector set from a template, and the template to pick is ‘System Performance’. You can then click on ‘next’ to further customize the set, or flick ‘Finish’.
Customize the Data Collector Set
There are some further customizations which should be done to the data collector set which was just created. In order to do that locate it in the Performance Monitor ‘User Defined’ data collector sets, and select it. The right hand window will now show the two components of the data collector set.  It is not normally necessary to capture the kernel information, so right click on that and choose ‘Delete’.
Then right click on ‘Performance Counter’ and choose ‘Properties’.
You can review the data which will be captured on the ‘Performance Counters’ tab.
The main thing to adjust is the sample interval. This is how frequently data is captured. The default is 1 second. This will generate very large files which are difficult to manipulate and share amongst customers, partners and Quadrotech Support.
It is suggested that if the data capture is going to run for a few hours that the sample interval is set to 15 seconds.
If the data capture is going to run for 24 hours or longer, it is suggested that the sample interval is set to 1 minute.
Run the Data Collector Set
Once the customization of the data collector set has been performed, the collector can then be run. To do this navigate to the ‘User Defined’ data collector set that was created earlier.  In the left hand pane, right click on the Data Collector Set, and edit the properties.  The default is that the collector will only run for 1 minute. This needs to be changed on the ‘Stop Condition’ tab. Depending on the requirements of the data collection this can be configured so that it will collect data until manually stopped, or, it can be configured to capture data for a number of hours.

Once configured, you can right click on the data collector and choose ‘Start’:

The icon for the data collector will then change and show that it is running (as shown in the screenshot above)
Reproduce the load or other condition
Once the data collector has been customized and is running, the server can then be used to perform the required operations or the normal server load reintroduced.
Review the captured data
Once the data capture has run for the required period of time, the capture can be stopped (if it wasn’t previously configured to stop after a known period of time). Once stopped, navigate to the folder where the data was capture and check that a file was created that seems to be a reasonable size and therefore is likely to be usable. The folder where the data has been collected is shown on the collector. For example:

As a general guideline you can expect about 10 Mb of data to be collected per hour, when collecting at 15 second intervals.

Print Friendly, PDF & Email