After you have selected a source MUF, decide on the schedule of snapshots that will best serve your needs.
For first time users of the AutoCollect system, we recommend that you take either a weekly or at most a daily snapshot.
As you become more experienced, you may want to do additional snapshots to get more granular information. However, the more snapshots taken, the more data that is collected and needs to be managed.
Example 1
You want to build a set of weekly statistics to compare weekly activities and to evaluate various tuning changes. This is the recommended schedule for the beginner doing Business Value Metrics (BVM) measurements and tuning.
For this example, the solution is to insert one additional step in front of the MUF EOJ step that requests the snapshot to be taken. If the snapshot fails for some reason, the MUFEOJ step should be skipped. This allows the MUF to stay up and for the snapshot attempt to be re-issued to verify that the data is collected before the MUF is ended.
Example 2
You want to build a set of weekly statistics to compare weekly activities and to evaluate various tuning changes.
For this example, the solution is to require the operator to submit the Snapshot function batch job prior to issuing the EOJ command.
An alternative would be that in normal occurrences, the workload from Saturday at 17:00 until the MUF is cycled is not significant. You might decide to just have a scheduled snapshot that runs every Saturday at 17:00. This way you can compare each week as the period from when MUF comes up until 17:00 on Saturday and would not require special attention from the operator.
Example 3
You want to build a set of weekly statistics but MUF is cycled on a monthly, or on an as-needed basis. This means that it may not be predictable which day of the week the MUF is cycled.
For this example, the solution is to have an automated job that runs Saturday at 17:00 to create a snapshot. In addition, a final snapshot must be taken for the end of MUF statistics just prior to MUF EOJ.
Each week that includes a MUF cycle would actually have two or more snapshots that must be added together to get that week's statistics.
You can use the AUTOCOLL OPTION=SUMMARY function to combine these multiple rowset periods (for a given week) into a single new SUMMARY rowset providing the ability to combine the full week's data into a single rowset for comparison purposes. For information about the AUTOCOLL OPTION=SUMMARY function, see the CA Datacom/DB DBUTLTY Reference Guide.
Example 4
You want to build a set of weekly statistics, but also want to see daily activity between 9:00 and 17:00. This would allow you to compare important times each day and statistics.
For this example, the solution is to have an automated job that runs the Snapshot function each day at 9:00 and at 17:00. In addition, a final snapshot must be taken for the end of week total. This final snapshot is either at 24:00 on Saturday or just prior to MUF EOJ.
As stated in the preceding section, the first time users for AutoCollect should choose a simpler schedule of Snapshot functions to meet their needs and then add additional snapshots as they become more familiar with the system.
Example 5 (Complex Situation) – building two sets of statistics from one MUF
In certain complex environments, the MUFNAME parameter that was discussed earlier in regards to MUFPLEX (including Shadow MUF) environments can be used as a means to collect snapshot data with two different goals in mind.
You want to have weekly snapshots to use for weekly comparisons as suggested in Business Value Metrics and to have snapshots for critical processing done during the daylight hours of Monday through Friday.
In this case, consider using the OUTNAME parameter as a way to segregate the two sets of snapshots. The schedule should be as follows. The rows for Sunday through Friday are daily snapshots and the row for Saturday is the weekly snapshot. Because the snapshot rowsets always contain accumulation values you get the full statistics for the weekly snapshots even though there were intervening daily snapshots.
Sun 8/6 00:00 MUF is initiated Mon 8/7 09:00 Snapshot with OUTNAME PRODMUFD Mon 8/7 17:00 Snapshot with OUTNAME PRODMUFD Tue 8/8 09:00 Snapshot with OUTNAME PRODMUFD Tue 8/8 17:00 Snapshot with OUTNAME PRODMUFD Wed 8/9 09:00 Snapshot with OUTNAME PRODMUFD Wed 8/9 17:00 Snapshot with OUTNAME PRODMUFD Thu 8/10 09:00 Snapshot with OUTNAME PRODMUFD Thu 8/10 17:00 Snapshot with OUTNAME PRODMUFD Fri 8/11 09:00 Snapshot with OUTNAME PRODMUFD Fri 8/11 17:00 Snapshot with OUTNAME PRODMUFD Sat 8/12 24:00 Snapshot with OUTNAME PRODMUFW
AutoCollect provides the ability to process Snapshot rowsets by MUFNAME. This allows you to create spreadsheets and BVM data based on either the daily information for PRODMUFD or the weekly data for PRODMUFW.
Example 6 (Complex Situation) – building two sets of statistics from one MUF but storing the data separately in two repositories
Another alternative is to define two different MUF environments as repositories for a given source MUF. When executing the Snapshot function, you can point to the appropriate repository for that Snapshot rowset using DBSIDPR.
For example:
Each day snapshots are taken every three hours and directed to the MUFSYSP repository.
Once a week on Saturday at 24:00, weekly snapshots are taken and directed to the MUFPROD repository.
Note: The flexibility of the snapshot schedule coupled with the ability to assign MUFNAMEs and select different repository MUFs gives the user a wide choice of options for implementing AutoCollect. The previous six examples are just a few of the possible scenarios.
|
Copyright © 2014 CA.
All rights reserved.
|
|