IBM Support

Performance Graphs over multiple days - from R720 Collection Services Data

News


Abstract

This page covers tips on generating performance graphs over a period of more than one day using Navigator for i on IBM i 7.2 or systems without historical data.

Starting in IBM i 7.3, a new capability to generate a historical data collection was added. This collection summarizes Collection Services performance data over multiple days. Using this historical data, graphs of performance data can be charted over multiple days using the Graph History feature of Performance Data Investigator (PDI) within IBM Navigator for IBM® i. Note that this feature needs to be turned on for the creation and saving of Historical Data over time. See this article for more information: https://www.ibm.com/support/pages/node/6217189

However, if you have not upgraded to IBM® i 7.3 yet, there are still ways that you can view multiple days worth of performance data using the traditional Collection Services collections.

Content

You are in: IBM i Technology Updates > Performance Tools > Performance on the web > Tips and techniques > Performance Graphs over multiple days
Starting with IBM® i 7.3, Collection Services added historical data collections which help produce graphs over multiple days. If you are still on IBM® i 7.2, there are options available to find a great deal of data from Collection Services (CS) with the classic information available.

Strategies for Saving Performance Data
You can save specific collections over time by copying or moving the collections from the default collection library to another library.  Leaving collections in the default collection library will cause them to be deleted when they expire.  Another option is to set a permanent retention period on the management collection (*MGTCOL) objects and/or the standard database file (*CSFILE) collections. 
Once you have implemented a strategy to save collections, you will be able to view one or more days of performance data at a time and compare any data that is saved from a year ago, the past 5 years on Jan 1 or a certain date, etc.
To conserve the amount of disk space needed to save performance collections, you may decide to only save the *MGTCOL objects and then generate the performance database files when you want to graph the data from that time period.

Viewing multiple collections in Performance Data Investigator (PDI) charts
The latest release of PDI (2020 Q2) allows you to select multiple collections in a library for display.  Start by displaying a a graph for a single collection from the library you are interested in.  Next, open the context panel, select additional collections, and click 'Refresh'. 
Displaying multiple collections is no longer limited to a library with 5 collections or fewer.  You may select one or more collections in any library for the dates you are interested in displaying.  See details on viewing multiple collections at: 
Note that adding more collections does increase processing and display time.  Keep this in mind when you are choosing collections to display together. To help minimize this overhead, increase the graph interval size to plot fewer points over the multiple collections.
By selecting certain collections over a long period of time, you can do more analysis on trends for your system performance. 
By selecting collections from specific dates, you can compare the performance from year to year, month to month, or week to week.

Graph Interval
With the modernized charting for PDI first announced in 2019 Q4, there is now a selectable Graph Interval.  When selecting more than one collection, it is recommended that you verify and possibly increase the graph interval to display. Selecting a larger graph interval will save on processing and display time.
Keep in mind that the collected metrics which are not displayed with a longer graph interval using collection services database files (*CSFILE) are not summarized into the plotted data as they are for historical data (*HSTFILE) available in IBM® i 7.3 and later releases; they are skipped.  It is efficient because a smaller amount of data is processed.  But it also means that you could miss peaks and valleys that potentially could occur only in the skipped data. 
For example, if data is collected at 5 minute intervals, and you choose a 15 minute display interval, there will be 2 intervals of data (10 minutes) out of every 15 minutes that are not included in the charted data.  There will not be an average or maximum calculated for the interval when that interval selected is larger than the collection interval set when creating the database files (*CSFILE). 
This provides an efficient way to display metrics over more time and compare at a high level view.  Then you can display a one day collection to look for specific data within a certain time span and all data collected.

Collection Services Data
You also can be smarter about creating the Collection Services database files to help with efficient use of data over a longer period of time.  Using Create Performance Data in Navigator for i or the CRTPFRCOL command, you can create files that have a larger interval and also include all the collected metrics.  
By setting the interval parameter to 60 minutes (the largest available), all the data will be included for each metric saved.  This way, you will get the 60 minute graph interval by default without losing any information.  This would be useful for more efficient display of multiple collections.
For more information on configuring Collection Services collection time interval and creating Collection services database files, see Collection Services Collections.
How to Stop/Turn-Off PM Agent 
As of September 30, 2020, Performance Management for Power Systems (PM) is withdrawn as a standalone offering in addition to be removed from other premium support service offerings. 
If you have utilized Performance Management Agent (PM Agent) for automated collection of nonproprietary Collection Services data, or if you have had PM Agent turned on to allow longer collection of historical data, you may now select to turn PM Agent off.  This will not have any effect on the length of time that historical data will be saved on your partition.  These changes take affect starting with the Fall 2020 Navigator for i PTFs.
You may select to turn PM Agent off or leave it as it is.  
The following steps can be followed to stop PM completely on the server. 
This will prevent PMAgent data to be sent and disable the function:
1. From Command Line, type CFGPMAGT and press ENTER
2. Press ENTER on 'PM Agent Signoff Approval' page
3. Enter option '0 = do not send performance data' for 'Select connectivity option for sending performance data to IBM'
4. Enter option '0 = No' for 'Receive performance data' Press ENTER twice to return to the Command Line.
5. From Command Line, type GO PMAGT and press ENTER
6. Select option '2. Work with Automatically Scheduled Jobs'
7. Enter option '2=Change' before each Scheduled Job that is listed (Q1PTEST, Q1PPMCHK, Q1PPMSUB, Q1PDR, Q1PPG)
8. On the 'Change Automatically Scheduled Jobs' screen, change 'Status' to 'I=Inactive' and press ENTER. Press F12 to return to the Command Entry
9. From Command Line, type the following command to remove Q1PSCH job:
 CALL PGM(QSYS/Q1PENDPM) PARM('*IMMED ' X'999999' X'00000000')
Note: There must be four spaces after *IMMED and one space before each X.
10.  There is an autostart job entry for QPM400.  This will be removed in a future PTF.  To remove the QPM400 autostart job entry now, call this command:

> RMVAJE SBSD(QSYS/QSYSWRK) JOB(QPM400)

To select to no longer transmit data:
  1. Run PMCFG command
  2. Select to not transmit data

[{"Type":"MASTER","Line of Business":{"code":"LOB57","label":"Power"},"Business Unit":{"code":"BU058","label":"IBM Infrastructure w\/TPS"},"Product":{"code":"SWG60","label":"IBM i"},"Platform":[{"code":"PF012","label":"IBM i"}],"Version":"7.2.0"}]

Document Information

Modified date:
08 October 2020

UID

ibm16217188

Page Feedback