Example 2: Deep Drill Down Analysis with Queryband Parameters - Teradata Workload Analyzer

Teradata Workload Analyzer User Guide

Product
Teradata Workload Analyzer
Release Number
17.00
Published
September 2019
Language
English (United States)
Restricted Access
TTU-17.00-EAP
Last Update
2019-06-07
dita:mapPath
vjd1544831946946.ditamap
dita:ditavalPath
ft:empty
dita:id
B035-2514
Product Category
Teradata Tools and Utilities
This example describes deep drill-down analysis of several query band parameters to help identify and isolate various request clusters, or provide additional granularity on request clusters.

In this example, one initial workload consumes the majority of the resources. A more granular breakdown of that workload is investigated. Long running outliers are noted in the analysis. The goal is to have these outliers classified into their own workload so that different workload management techniques are applied.

  1. Select Analysis > New Workload Recommendations. The Define DBQL Inputs dialog box appears.
  2. Complete the Define DBQL Inputs options. For more information, see Defining DBQL Inputs.
  3. In the Category box, select Account String, and click OK. The Unassigned requests report appears.
  4. Right-click over each workload and select Auto-Generate Workloads. The Candidate Workloads Report appears.

    The CPU distribution of all workloads is 8% ADW-TACT and 92% ADW-DS. Workload WD-ADW-DS is selected for further analysis because it is consuming 92% of the total CPU.





  5. Right-click WD-ADW-DS, and select Analyze Workload. The Analyze Workload tab appears.


    There are a total of five distinct query band names. The query bands can be viewed only if Queryband is selected from the Workload Correlation Parameter list. The five distinct query band names in this example are:
    • QueryBand (5)
      • AggLevel (7)
      • Function (5)
      • Region (7)
      • TopTierApp (3)
      • Urgency (3)

    There are several suitable analysis candidates available in this query band list, as denoted by the distinct value counts.

  6. For this example, select QueryBand as the correlation parameter.
  7. Click to load the query band names in the Queryband Filter drop-down list.
  8. Select Function, then click Perform Analysis. The Correlation Report and graph appear.


    A total of five query band values display for the Queryband=Name function. Notice a possible distinction with Function=MIN, which included queries with greater time spent than any of the other queries.

  9. From the report, right-click the MIN row, and select Split to New Workload.
  10. Name the new workload WD_ADW_Outliers. This step is done to insure that unassigned clusters fall back into the original WD-ADW-DS workload classification.


    The QueryBand Function=MIN is split to the new workload, ADW-Outliers. The remaining four functions are unassigned, falling back to ADW-DS if no other action is taken on them.



    The Correlation/Distribution Reports and graph refresh with the remaining four unassigned query band values for the next add or split operation.