Real-Time Tools

In the z/OS mainframe environment the free tools are part of the RMF suite of products. These consist of RMF Monitor II, RMF Monitor III, and RMF PM.

RMF Monitor II is used for snap shot analysis. This monitor has two modes: interactive and historical. The interactive session is started by issuing command RMFSTART  in TSO and then selecting option 2. You will then be presented with a menu to choose what type of data to display. This primarily shows resource usage.

RMF Monitor III is for short term performance analysis.  This session is started by issuing command F RMF, S III  An address space called RMFGAT is then started that controls the collection of monitor III data. Please note that RMF must have already been started with the S RMF command.  The monitor III display is started by issuing command RMFSTART  in TSO and then selecting option 3. RMF Monitor III was originally called RMFWDM; which stands for RMF Workload Delay Monitor. Its major function is to show the resources that is causing a job to be delayed.

RMF PM
is a graphical enhancement for RMF online monitoring that allows you to analyze RMF Monitor III data right from your workstation. It requires an address space called GPMSERVE to be active on the host.

For more information on RMF check out the following web sites:
http://www-03.ibm.com/servers/eserver/zseries/zos/rmf/
http://www-03.ibm.com/servers/eserver/zseries/zos/rmf/Library/

Many large organizations also purchase additional tools. The most prevalent of these are OMEGAMON (IBM),  and TMON (ASG). In addition to monitoring resource usage they have facilities to look at z/OS control blocks, issue commands, and determine the delay reasons for individual jobs; which, along with other features, provide added value over some of e free tools.

For more information on OMEGAMON check out the following web site:
http://www-01.ibm.com/software/tivoli/products/omegamon-xe-zos/

For more information on TMON check out the following web site:
http://www.asg.com/products/product_details.asp?code=TMV

In contrast to the z/OS world the UNIX/LINUX environment has a more diverse set of free tools. Due to the fact that there are different flavors of UNIX (e.g., LINUX, AIX, HPUX, SOLARIS, etc.) there are usually subtle differences between the use and output from the various tools.

The most often used tool in the UNIX/LINUX world is SAR (System Activity Reporter). This is both a historical reporting tool and a near real-time tool. It is near real-time because it shows a snapshot of the system resources averaged over a specific interval (e.g., 5 or 10 minutes) and not what is necesarily occurring at that moment. SAR is started at boot time and the reports are accessed by issuing the command SAR.

The most often used real time tools are TOPAS and TOP. Most versions of UNIX (including LINUX) use TOP. On AIX TOPAS is used. They are very similar in that they show a snapshot of system resource usage refreshed over a specific interval (usually 5-10 seonds).

For more information on TOP you can enter man top on your server or check out the following web site:
http://www.unixtop.org/man.shtml


For more information on TOPAS you can enter man topas on your server or check out the following web site:
http://publib.boulder.ibm.com/infocenter/systems/index.jsp?topic=/com.ibm.aix.cmds/doc/aixcmds5/topas.htm

IBM also provides a free tool for the AIX and LINUX environments called NMON. NMON was originally developed by IBM as a free unsupported tool for AIX. They then extended it to the LINUX environment and as of AIX 5.3 TL09 it is bundled with AIX and is fully supported. This version is called TOPAS_NMON.

For more information on NMON check out the following web sites:
http://www.ibm.com/developerworks/aix/library/au-analyze_aix/
http://www.ibm.com/developerworks/wikis/display/WikiPtype/nmon

In addition to TOPAS, TOP and NMON you can also use VMSTAT and IOSTAT for real-time analysis.

The VMSTAT command reports virtual memory activity. This includes memory usage, paging, and CPU utilization. 
Entering the VMSTAT command without any flags displays a summary of the statistics since system startup. In order to use VMSTAT as real-time tool you must specify the amount of time between each report and the number of reports to be generated. Please note that in most cases a report is one line of data. Once VMSTAT has produced the number of reports requested it stops and must be restarted in order to use it again.VMSTAT is one of the commands that have differences between the various versions of UNIX so you have to check the man page (man VMSTAT) to determine the available flags for your version of UNIX. For example: In AIX if you specify the -t flag the timestamp will be displayed next to each line of output. In LINUX this feature is not available.

Click here for an example of  AIX VMSTAT output
Click here for an example of LINUX VMSTAT output

The IOSTAT command reports I/O activity.  Entering the IOSTAT command without any flags provides some basic information per device such as IO's per second. It also provides CPU information. If you want more detailed information per device such as average service time you need to specify the -D flag in AIX and -x flag in LINUX. In order to use IOSTAT as a real-time tool you must specify the amount of time between each report and the number of reports to be generated as with VMSTAT. Once IOSTAT has produced the number of reports requested it stops and must be restarted in order to use it again.  IOSTAT is one of the commands that have differences between the various versions of UNIX so you have to check the man page (man IOSTAT) to determine the available flags for your version of UNIX.

Click here for an example of AIX IOSTAT output
Click here for an example of LINUX IOSTAT output


For more information on VMSTAT or IOSTAT you can enter man vmstat or man iostat on your server or check out the following web sites for the LINUX version of the command:
http://linuxcommand.org/man_pages/vmstat8.html
http://linuxcommand.org/man_pages/iostat1.html

In addition to the free tools mentioned above there are also additional tools that can be purchased for a fee similar to the z/OS mainframe. The one that I am familiar with is called GlancePlus by HP. This is supported on most of the flavors of UNIX (including LINUX) and provides more information and a better interface than the tools mentioned above. BMC also has a product called PATROL that I believe provides similiar functionality; although I have never used it.

For more information on GlancePlus check out the following website:
https://h10078.www1.hp.com/cda/hpms/display/main/hpms_content.jsp?zn=bto&cp=1-11-15-28%5E9637_4000_100

Please click here to sign my guestbook



All comments and questions are appreciated. Please contact Joel Wolpert at joel@perfconsultant.com


Performance Consultant Logo