The use and analysis of VIOS Health Advisor tool

Introduction to VIOS Health Advisor tool

The VIOS Health Advisor tool is VIOS2.2.2.1 and does not need to be installed separately. It provides the overall system health report for VIOS after a period of time (at least 10 minutes, up to 1 hours) of the VIOS system, and can analyze the performance bottleneck of VIOS system, and give some suggestions to solve the performance problem.

Any long-running monitoring tool has overhead, VIOS Health Advisor is no exception, but VIOS performance Advisor consumes less system resources. If the system is very busy, such as physical CPU idle less than 0.2 Core, then do not recommend the collection of Health Advisor information, but can collect nmon information, and then use Health Advisor analysis. If the Health Advisor information is still collected while the system is busy, because the tool does not proactively preempt the CPU resources used by the application, the result is that the command waits for the system to be idle before the report is finally generated.

VIOS 2.2.2.1 Health Advisor Performance monitoring

The Health Advisor tool command, which needs to be executed in a VIOS environment, is part (performance analysis & Reporting Tool). The execution syntax is specific as follows:

$ part
Usage:part {i INTERVAL |-f FILENAME} [t level] [--help|-?]
       -I <minutes>    interval can range between 10-60-
       f <file> any       Nmon recording- 
       t <level>      1-basic logging, 2-detailed logging
       -?              Usage message

The-I parameter sets the time for monitoring, ranging from 10 minutes to 1 hours.

The-f parameter is to parse the result of the Nmon, and the resulting Nmon file is appended to the parameter.

-T parameter to set the verbosity of the analysis, the information that is collected by default is the Detailed logging level.

CPU Utilization Monitoring

First, test the profile file configuration of the partition VIOS:

Figure 1 View the partition profile

Under initial conditions, the system has no pressure to collect 10 minutes of system performance data.

$ part-i 10part:reports are successfully generated in 730-13-vios1_121203_01_09_19.tar

After 10 minutes, the command completes. Get the. Tar compression pack to the Windows machine, unzip it, and then open the file in the Browse area: Vios_advisor_report.xml

Focus on CPU part information on the page:

Figure 2 View the CPU part information in the analysis results

In several of the icons in the previous illustration, the White hook icon indicates that this is normal and does not need to be adjusted.

White exclamation mark icon needs attention, namely informative. The Risk and Impact levels are both 5 highest and 1 lowest.

Because the CPU is not under any pressure during this test, the value of "suggested value" is null in the CPU metrics, without adjustment.

Next, use the NCPU command, poll the CPU for pressure, and then use VIOS performance Advisor to collect 10 minutes of information:

The partition has 16 logical CPUs and starts 15 NCPU processes:

#/ncpu-p
/ncpu-processes=15 snooze=0% hibernate=0 secs Forever
# lparstat 1
System Configuration:typ e=Shared mode=Uncapped smt=4 lcpu=16 mem=4096MB psize=8 ent=2.00
%user  %sys  %wait  %idle physc %entc  lbusy   app  VCSW phint
--------------------------------------   -------------
  0.0   0.3    0.0   99.7  0.01   0.7    0.0  7.99   588     0 
  0.0   0.3    0.0   99.7  0.01   0.7    0.0  7.99   652     0    95.6  0.1 0.0 4.3 3.63 181.7   82.6  4.21  1389     0 
 99.9   0.1 0.0 0.0  4.00 199.9  100.0  4.00  1599     0 
 99.9   0.1    0.0    0.0  4.00 199.9  100.0  4.00  1599     0

Using the Lparstat command, observe the "PHYSC" column of values, the initial physical CPU is 0.01, after starting the NCPU process, the partition will acquire 4 physical CPUs. This is because the partition is a uncapped partition and the expected virtual processor number is 4.

Use Topas to view CPU utilization:

Figure 3 Nmon Viewing CPU utilization