Packages & Perspectives

Some charts are not available on the Investigate Data page.  They can only be displayed by selecting data from a primary perspective and then selecting a drilldown chart from the "Actions" button (found above the context panel).  These drilldown perspectives are referred to as secondary charts.

The primary perspectives display metrics (CPU, page faults, waits, etc) by a single category type (job or task, system, user profile, server type, etc).  Primary perspectives are typically named "[metric] by [category]".

Drilldown charts allow the user to select a specific category value to filter only those desired categories.  The secondary perspectives are typically the Perspectives with "for" in the name indicating they are a specific metric (CPU, Page Fault, Waits) for a specific subset (job or task, thread or task, pool, user profile, disk, subsystem, server type, job priority) with the format "[metric] for [category]" or "[metric] for one [category]".  This could be for more than one pool, job, etc, or it may specify "for one". 

An example of a primary perspective would be "CPU Utilization by Thread or Tasks", which displays the CPU utilization (metric) of threads and tasks (category) with the highest percentage during the collection. Users can then click on as many specific threads or tasks as they want and navigate to the "CPU Utilization for Threads or Tasks" drilldown graph. The chosen threads and tasks are now individual series plotted on a stacked bar chart against each other by time. The user can further drilldown by choosing two bars of the same thread or task and navigating to "CPU Utilization for One Thread or Task" to view the CPU of only that thread during that period of time. The primary perspective shows the total impact of each thread by the end of the collection, whereas the drilldown chart shows each thread's impact at every time interval during the collection.

Another aspect of the filtering that can be done on drilldown secondary perspectives is by time.  Selecting a metric on one interval will filter by that time interval.  Selecting the same metric on multiple intervals will filter by the time between the smallest and largest intervals chosen. 

Drilldown - required selection

Secondary perspectives require selection before drilling down from a primary perspective.  

Chart requested is missing fields for the correct results. Select a data point on the previous chart to provide the required context information.

Drilldown with multiple selections

The ones we are discussing for this section are not the "for one" perspectives which only allow (or adhere to) a single selection.  With multiple selections, there were potential visual issues with secondary charts that were fixed in Q3 2025.  

When multiple category entries are selected (threads, tasks, jobs, pools, etc).  In cases where the selected category entries did all not contain data at the same time intervals, the x-axis labels would become skewed and not line up with the display.  Additionally, if some selected categories contained more data than others (as a result of existing in the collection for more intervals than others), some intervals might be skipped or plotted on the wrong point on the x-axis.  When this occurs, the data in the table is correct, but the chart view is misleading.  You may notice this by the hover text interval data not matching the x-axis where that data is plotted.

This could happen when a thread started part way through the collection cycle, or ended before the collection cycle was completed, but other threads (also selected for drilldown) existed across the full collection.  

One way this may manifest is if you are on a "by" chart (CPU Utilization by Thread or Task), you select multiple entities (thread, task, etc), and drilldown to a "for" version of that chart (CPU Utilization for Threads or Tasks).  A breakdown dimension is used to show each of the entities, and if the selected entities do not all have metrics for the same intervals where data is charted, the values may not be correctly lined up with the x-axis labels.  This can be noticed if you hover over a data point (bar or line) and view the date/time stamp and compare with the value on the x-axis label for the date/time stamp.

Here are examples of perspectives that this may happen for from Collection Services package:

• CPU Utilization for Jobs or Tasks
  • for Threads or Tasks
  • for Generic Jobs or Tasks
  • for Job User Profiles
  • for Job Current User Profiles
  • for Subsystems (from CPU Utilization by Subsystem)
  • for Server Types
  • for Pools
• Physical Disk I/O for Jobs or Tasks (from Job Statistics by Job or Task or other)
  • for Threads or Tasks - Advanced, Basic, Detailed
  • for Generic Jobs or Tasks - Advanced, Basic, Detailed
  • for Job User Profiles - Advanced, Basic, Detailed
  • for Job Current User Profiles - Advanced, Basic, Detailed
  • for Subsystems - Advanced, Basic, Detailed
  • for Server Types - Advanced, Basic, Detailed
• Page Faults for Jobs or Tasks (drilldown from Job Statistics by Job or Task or other)
  • for Threads or Tasks
  • for Generic Jobs or Tasks
  • for Job User Profiles
  • for Job Current User Profiles
  • for Subsystems
  • for Server Types
  • for Pools
• Logical Database I/O for Jobs
  • for Generic Jobs
  • for Job User Profiles
  • for Job Current User Profiles
  • for Subsystems
  • for Server Types
• 5250 Display Transactions for Jobs
  • for Generic Jobs
  • for Job User Profiles
  • for Job Current User Profiles
  • for Subsystems
• Disk Details for IOPs
  • for Adapters
  • for Disk Pools
  • for Disk Units
  • for Disk Paths
• Disk Overview for IOPs
  • for Adapters
  • for Disk Pools
  • for Disk Units
  • for Disk Paths
• Database package perspectives:
  • Physical Database I/O for Jobs or Task - Basic, Detailed
    • for Threads or Tasks - Basic, Detailed (drilldown from Physical Database I/O by Thread or Task - Basic)
    • for Generic Jobs or Tasks - Basic, Detailed
    • for Job User Profiles - Basic, Detailed
    • for Job Current User Profiles - Basic, Detailed
    • for Subsystems - Basic, Detailed
    • for Server Types - Basic, Detailed
  • Logical Database I/O for Jobs -Detailed
    • for Multiple Generic Jobs - Detailed
    • for Multiple  Job User Profiles - Detailed
    • for Multiple Job Current User Profiles - Detailed
    • for Multiple Subsystems - Detailed
    • for Multiple Server Types - Detailed
  • SQL Cursor and Native DB Opens for Jobs
    • for Generic Jobs
    • for Job Current User Profiles

• For PDI, we removed the use of INNER in some of our queries to fix this problem. 
• Another approach is to change it to "INNER" (with delimiters) or change to some other name.
• To have the query work on a system that doesn't have the enhancement that caused this change, you might need to leave something at this spot in the query.  Changing OUTER to simply O works on all releases.

Additional Perspectives (PML):

1. In March 2020 to fix the drilldowns from the first two perspectives
2. In Jan 2020 for the Db2 changes that required the "OUTER" keyword to be changed to "O".