It has been concluded that the optimal working level of most IT components or even complete installations is roughly 60 percent.  When systems or separate components exceed this amount problems start to materialize. As a system administrator you need to be aware when problems may start and when upgrades are due.  It is much better to have a plan in place to monitor and identify when these upgrades will occur then to be suddenly hit by them.

 

We will look at a few components to high light what causes these stresses and identify ways to measure the rate of change.  Each example requires that measures are collated and recorded.  With the use of a charting package future predictions are be projected. 

 

To accurately project when upgrades are needed the captured data has to collected at the right time of the day. We need to identify the maximum stress periods.  For example if we are measuring network capacity we need to record the levels at a time when the network experiences its maximum usage.  This would more than likely be early in the working day when users are logging on to the network and checking their email then it would be in the late afternoon. 

 

 

Measuring Component Usage

 

HDD Capacity

The more data that is stored on a hard drive, the long it takes the drive to locate it. Data is firstly stored on the out tracks and once these are filled, it is moved inwards towards the centre.  Locating data from the inner tracks places greater stress on the magnetic read write arms as it has to move over a greater number of smaller tracks to locate the clusters.

CPU Ultilization

The Performance Monitor would be a tool to use to record this type of data.  When the load reaches a problem level the CPU can be upgraded or an additional CPU added if faster CPU's are not available.

Memory/Paging File

Again the Performance Monitor can be used.  It may be an idea to measure how often the page file is used to compensate for lack of RAM or the size of the paging file used.