A few weeks ago I wrote about the various Performance Data Collectors in IBM i. It’s great to be able to collect such a wide variety of performance data, but what do you do when you want to manage that data? Maybe you have the need to do some analysis on a particular performance data collection, but you do not want to run the analysis tools (which have overhead) on your production partition. How do you move that collection to a different partition? Or maybe you are trying to develop a strategy for how to save (and possibly restore) your performance data; you never know when you might need a performance collection from a critical time period for comparison purposes if something goes wrong. Perhaps you simply want an easy way to delete some performance data that you no longer need.
Consider what a performance collection consists of: it may be data in a management collection object or it could be data in a set of database files. When the data is in the database files, each type of performance collection has its own set of files:
- QAPMxxx for Collection Services
- QAPYJWxxx for Job Watcher
- QAPYDWxxx for Disk Watcher
- QAYPExxx for Performance Explorer
Each collection consists of a member with that collection name in each of the files for that collection type. Prior to 6.1, it was rather difficult to manage an individual collection of performance data.
In the 6.1 release, in addition to adding the Job Watcher and Disk Watcher performance collectors to the operating system, IBM also made it much easier to manage all of your performance data. The blog, i Can Find All my Performance Collections in One Place, described the Collection Manager – the graphical user interface for managing all of your performance data collections.
That blog made brief mention of the capability to copy, delete, save and restore performance data through the GUI.
The support behind the GUI is actually a set of commands that were added to the 6.1 release. These commands are:
- Copy Performance Collection (CPYPFRCOL)
- Delete Performance Collection (DLTPFRCOL)
- Restore Performance Collection (RSTPFRCOL)
- Save Performance Collection (SAVPFRCOL)
All of these commands have a very similar interface and all of these commands can be used to manage any of the types of i performance data. You have to know the name of the performance collection you want to work with and the library in which it resides; you also have to specify the type of performance data you want to manage. You can select from the following (the default is *CSFILE):
Collection Services file-based collections
Collection Services object-based collections
Disk Watcher file-based collections
Job Watcher file-based collections
Performance Explorer file-based collections
Performance Explorer object-based collections
All types of performance collections. This includes file-based collections and object-based collections.
A bit of explanation is in order for “file-based” collections versus “object-based” collections.
When using Collection Services and Performance Explorer, data is initially deposited into a high-performing, highly scalability repository called a Management Collection object. The performance data can also be copied into a set of database files, which is required for analysis. The data in the management collection object is referred to as an “object-based” collection, while the data that has been copied into the database files is referred to as a “file-based” collection. You will note that for Disk Watcher and Job Watcher, no intermediate repository is used (the data is deposited directly into the database files), and as such, those types of collections only have options for “file-based” collections.