BMC Perf Mgr — Accessing Raw Data

Preparing BMC Perf Mgr Data for SAS IT Resource Management

BMC Performance Manager for Servers data is collected by BMC Performance Manager, which was formerly BMC Patrol. BMC Perf Mgr data can be collected from the following operating systems:

Windows NT
UNIX
SAP ERP

The BMC Perf Mgr Agent maintains parameter history data about the systems that it monitors. For more information, see your BMC Perf Mgr documentation.

BMC Perf Mgr enables each metric to be sampled at its own interval. This interval can be set by the BMC Perf Mgr administrator. SAS IT Resource Management requires that the sample rates be specified on minute boundaries. (However, you can also specify a sample rate of 30 seconds.)

SAS IT Resource Management provides support for three Knowledge Modules:

Windows NT Knowledge Module
UNIX Knowledge Module
SAP ERP Knowledge Module Version 2.2

For information about the performance areas that are collected by BMC Perf Mgr, see BMC Perf Mgr Adapter - BMC Perf Mgr Domain Category.

Extracting Data from BMC Perf Mgr

A raw data file of BMC Perf Mgr history data can be generated in two ways:

extracting data from the BMC Perf Mgr History Knowledge Module

The history knowledge module organizes the collection of the history data and ensures that it is sent to a central server from which it can be extracted. For more information about this method, see the documentation from BMC Perf Mgr about the History Loader Knowledge Module.
extracting data by using the dump_hist.exe command

This command extracts the same history data that is extracted from the History Knowledge Module. However, it does not manage the transferal of the data to a central location. (This option is useful if you prefer writing your own scripts to control the extraction and transferal of the data to a central location.)

Note: When transferring the history data across platforms, ensure that the data retains its ASCII text format.

Although these two methods produce slightly different output, both can be processed by the staging transformation of SAS IT Resource Management. The BMC Perf Mgr Operator Console retrieves the historical data that is stored by the Agent. The dump_hist line command copies the parameter history data that is maintained by the BMC Perf Mgr Agents. The BMC Perf Mgr Agent Reference Manual contains more detailed information about the dump_hist command. (The dump_hist.exe program is provided as part of the BMC Performance Manager software.)

The following command copies parameter history data for one day for the file that is indicated by filename. The command uses start (-s) and end (-e) parameters to specify the date and time of data that is to be copied. The format of these parameters is MMddhhmm[yyyy], where

MM= month
dd= day
hh= hour
mm= minute
yyyy= year, which is optional

The following code copies the full day of data from July 18, 2007.

-  dump_hist -s 071800002007 -e 071823592007 > filename

The following example shows the format of the text file that is created by the preceding dump_hist command. This file can be input to the staging transformation of SAS IT Resource Management.

     machinename/NT_CPU.CPU_0/CPUprcrUserTimePercent
        Thu Jul 18 10:00:57 2007 26.981
        Thu Jul 18 10:01:58 2007 5.35963
        Thu Jul 18 10:02:58 2007 0.598205
        Thu Jul 18 10:03:58 2007 0.333915
     machinename/NT_CPU.CPU_0/CPUprcrPrivTimePercent
        Thu Jul 18 10:00:57 2007 61.0279
        Thu Jul 18 10:01:58 2007 1.20528
        Thu Jul 18 10:02:58 2007 1.56053
        Thu Jul 18 10:03:58 2007 1.05312
     machinename/NT_SYSTEM.NT_SYSTEM/SYSsysTotalProcTimePercent
        Thu Jul 18 10:00:57 2007 88.013
        Thu Jul 18 10:01:58 2007 6.56211
        Thu Jul 18 10:02:58 2007 2.1812
        Thu Jul 18 10:03:58 2007 1.36592

To dump data for a single server, add the —host<hostname> parameter to the previous command. The command then dumps a full day of data for the single server that is identified by <hostname>.

For additional information about the syntax of the dump_hist, see the PATROL® Script Language Reference Manual.

Notes about BMC Perf Mgr

When staging BMC Perf Mgr data on z/OS, allocate the z/OS file with an LRECL= 200. (BMC Perf Mgr data typically has variable length records, but they do not exceed 200 bytes in length.)

Processing BMC Perf Mgr history data into a staged table might require the following two issues to be resolved:

different sample rates are collected for each metric

Two metrics 'A' and 'B' are not necessarily sampled at the same rate. Metric 'A' might be sampled at 1-minute intervals, and 'B' might be sampled at 5-minute intervals. To combine these two metrics into the same observation in a table is invalid because each value should eventually be weighted by the duration variable. To resolve this problem, the staging code of SAS IT Resource Management includes a variable in each BMC Perf Mgr table called DURGRP. DURGRP is a string that represents the duration group to which a metric belongs. In this example, metric 'A,' which is sampled every minute, is included in the observation with a DURGRP value of 60 (60 seconds). Metric 'B' is included in an observation with a DURGRP of 300 (300 seconds).

The DURGRP variable is used only during staging to ensure that the metrics are reduced and summarized by their respective duration values (assuming that they are weighted by DURATION).

If numerous null values are present in each observation, then the BMC Perf Mgr data in the SAS IT Resource Management tables might look unusual. The number of DURGRPs and null values depends on the number of different sample rates that are applied to metrics that belong to the same table.
Datetime Stamps of Samples Are Not Exactly Aligned
In this example, two metrics 'A' and 'B' are both sampled at 1-minute intervals. In the following example history data, the first sample occurred at x for both metrics. However, in the second sample, the datetime stamps differ by a second. That is, 'B' is sampled later than 'A'. The first sample for each metric is combined into a single observation because the duration and datetime stamps are the same. However, this is not the case for the second sample.
```
     machinename/NT_CPU.CPU_0/A
        Thu Jul 18 10:00:57 2007 26.981       
        Thu Jul 18 10:01:58 2007 5.35963 
     machinename/NT_CPU.CPU_0/B 
        Thu Jul 18 10:00:57 2007 61.0279 
        Thu Jul 18 10:01:57 2007 1.20528  
```
During the staging of the raw data, SAS IT Resource Management detects that this second sample has datetime values that are related. It then collapses the data into one observation. The result is that the data in the staged table is less sparse. However, the datetime and duration values are only near approximations.