Software User's Guide
The SPS/Performance Monitors

Introduction
list_users
Collecting system meters and creating reports

Introduction

The SPS/Performance Monitors address all aspects of system performance: capacity planning, current and historical data reporting, quick on-line troubleshooting and system-level alert handling. SPS monitors help identify which process or application is "killing" the system and resolve potential problems ahead of time to allow more accurate capacity planning.

SPS provides an on-line graphic representation of system's resource consumption. SPS captures and calculates system utilization data based on configurable combinations of groups, users and processes. The product is user-friendly, its reports and graphs are easy to interpret.

The reporting facility allows the user to examine historical data in On-line Playback Mode and to download report files into spreadsheet applications. SPS also features a system wide Analyze System Server process that reduces the overhead associated with the analyze_system subsystem and allows users to extract periodically and log any set of benchmark values.

SPS produces alert signals and automatically executes corrective commands when any of the monitored meters exceed the allowed thresholds.

You can meet these objectives with SPS/Performance:

Determine the performance characteristics of my application.
Determine the maximum user load capacity.
Demonstrate application and system scalability.
Demonstrate the application meets the performance requirements.
Determine application and system break point.
Determine if the current infrastructure can support the application.
Ensure that the infrastructure configuration meets requirements.
Ensure that the system operate at with tolerable response times at peak hours.
Ensure that the system can sustain predefined throughput.
Ensure that the system is optimally configured for performance.
Evaluate impact of application on current infrastructure.
Understand system behavior with different transaction mix.
Pinpoint and remove application and system bottlenecks.
Know you limits and prepare for future capacities.

Features

Records system utilization meters: CPU utilization, page faulting time and rate, disk I/O and system interrupt rates.
Monitors transaction rate and identifies processing bottlenecks.
Analyzes historical data (hourly, daily, weekly and monthly) and creates spreadsheet-ready, ASCII report files that can be imported into EXCEL.
Monitors application queuing activity, messages processed, messages pending, queuing rate and queue depth.
Features a real-time, graphical view of all monitored modules.
Provides process listings (list_users) sorted by: CPU utilization, I/O rate, Page faults and Interrupt rates.
Supports multi-module, multi-system sites.

list_users

The sps_lui.cm command is is an interactive tool that is used to monitor running processes. The key feature is that the program sorts all running processes by any performance meter - cpu, page-faults, I/O rate.

List-Users sorted by CPU Usage:

List-Users sorted by Page-Faults

List-Users sorted by Disk-Writes

Collecting system meters and creating reports

To start the collection process, execute the start_performance_server.cm command. This will start the SPS/Performance_Server process. Data is collected at one minute intervals 24x7 into daily files under the logs directory - logs>module_data.yy-mm-dd. The following information is collected:

Module Data

%CPU utilization
%Page faults
%Interrupts
%Empty-idle
File I/O rate
Disk I/O rate
%System memory used
%user memory used
%cache memory used
%free memory
Paging space utilization
Transactions processed
Transactions processing rate

Disk Data

Blocks left
%Used
Read operations / second
Write operations / second
%Read-busy
%Write-busy
Total I/O rate

Queue Data

Messages on-queue
Messages pending
Queue Highest count
Queue total messages processed
Messages per second rate

Process Data (top 5 processes)

Process name
User name
Program name
%CPU used over the last one minute interval

To create comma-delimited (csv) files for further Excel processing can be produced by running the performance_server.pm program.

performance_server.pm

Purpose

The command performance_server.pm with the -report argument creates CSV (comma-delimited) formatted reports in the logs sub-directory.

CRT Form

------------------------------ Performance Server -----------------------------
-report:        yes
-from:          07-05-26_00:00:00_EST
-to:            07-05-26_23:59:59_EST
-resolution:    1
-report_name    m1
-scaling_factor:
-max_processes: 2048

Lineal Form

performance_server.pm  [-report]  [-from]  [-to]  [–resolution]  [-report_name]  [-scaling_factor]  [-max_processes]

Arguments

-report

Cycle to YES to produce the report files. Three reports files will be created in the logs directory, all names prefixed with the -report_name value.

Module data: module_report.csv
Disk data: disk_report.csv
Queue data: queue_report.csv
Process data: process_report.csv

-from .... -to

Two standard-VOS data-time fields that define the time range of the report. Be default the system selects the current day.

-resolution

A number that defines the resolution of the reports in minutes. A resolution of 1 (which is the minimum) means that ever line represents 1 minute. A report with a resolution of 60 will calculate one line for each hour and so on. Note that you can use Process-data only with a resolution of one minute.

-report_name

A prefix to all CSV report files that will help to identify the reports.

-scaling_factor

The scaling factor to be used on V-Series modules. For more information see VOS documentation on display_system_usage and how to adjust the scaling factor for hyper-threaded processors.

-max_processes

The maximum number of processes running on the system.

Table of contents