Interpreting the Data 

Definition

If the monitors display N/A, the data is not available on that platform. Earlier R/3 Releases (up to Release 2.2D) often display a 0 or -1 instead of N/A to show that the data is invalid.

CPU (process management)

For each CPU, the following data is displayed:

These values show how much CPU is being used as it is being conveyed to the UNIX or vmstat . Some operating systems also display a wait value. From R/3 Release 3.0C, the wait values will be separate from the idle value. You can determine from this data whether any CPU resources are still available.

A process could loop if a CPU user is permanently busy.

Further values include:

Load average is the average number of processes per CPU which have been waiting to be assigned to a free CPU for over 1, 5 or 15 minutes. If the average is 1 process for each free CPU, then no problem has occurred. However, if the average is more than 3, there is a bottleneck in the CPU resources.

Top CPU Consumers

Memory

Paging

For most operating systems, the page out rates are more important than Page in rates. A Page in also takes place when a program is started, and does not necessarily lead to performance problems. However, if the actual Virtual Memory Management must store a page in the swap space several times (Pages out), then there is a memory bottleneck which significantly reduces performance. Which Pages out rates are critical depends on the current operating system.

Swap space

The most important details displayed in the Swap section are the Free in swap space and the Actual swap space.

You can provide the virtual memory management with swap space in various ways, depending on the operating system's configuration. Firstly, whole partitions or disks can be reserved for the swap, and this space is then marked as Configured. Swap space can also be a dynamically increasing file in a file system, and its values are displayed in Current. As the size of the file can increase, remember the limits of the swap space. All types of swap space are not available on all operating systems. Therefore, in some cases the actual size of the swap space must correspond to the configured and maximum swap space.

Disk

The average response time is also displayed in the Operating System Monitor. This is the sum of the wait time and the service time. Some of the values are redundant, for example, the average wait time cannot be calculated from the number of operations and average queue length. Note that all values have been included here, even though they are not all available on some platforms.

File System

LAN

Some values are not given in some network interfaces, for example, there may not be any 'collisions per second' in a token ring architecture. The values given here do not describe the actual network traffic. It describes the transfers made via this interface, for example, the errors displayed here are to do with the interface and not the actual network segment.

You can also view the mean values of data collected by the hour during the last 24 hours. The values under 'Recent hour' are taken directly from the shared memory. saposcol automatically determines whether or not the data is complete, and does not display incomplete data.

In addition, the background job SAP_COLLECTOR_FOR_PERFORMANCE reads the data from the shared memory, and then stores it in the performance database MONI .

 

See also:

Special Features of Operating Systems