File sizing introduction
Overview: two approaches to file sizing
After designing the data structures you are implementing (see Field design and Field group design) there are two ways for a file manager to approach the calculation of file sizes:
You can take the ad-hoc approach, by making sizing estimates and doing one or a combination of these:
- Iteratively load a sampling of records to verify
- Use the development and testing process to make final sizing decisions
Alternatively, you can do a detailed analysis of the data you expect the file to contain, and try to derive precise sizes for the Model 204 tables, a laborious process.
Choosing an approach
Most Model 204 file managers use the ad-hoc design approach. Often there is already a production file with similar characteristics to the new file you are creating. Simply copying its parameters as a starting point is a quick way to get a file ready for development and testing.
Moreover, most of the sizes can be easily changed dynamically, so an extreme level of precision is not overly important.
However, it is valuable for a file manager to be grounded in the principles of Model 204 file size calculation. To use (or just understand) the sizing rules, see File size calculation in detail.
Critical, up-front decisions
Because certain decisions are more difficult to fix, they must be as "correct" as possible, as early as possible. These are described in the sections that follow.
ATRPG and ASTRPPG parameters
Because the Internal File Dictionary is hashed, it can only be resized by reorganizing/recreating the file. This involves an outage of the Online, so it should be avoided.
You can look at the detailed calculation rules at Sizing Table A or, given how small Table A is compared to the other tables, round up on your rough estimate for ATRPG and round down (to try and fit fewer field definitions on each Table A page) for ASTRPPG.
CSIZE parameter
Like the internal file dictionary, Table C is hashed and so cannot be dynamically changed.
The easiest way to make sure that this is an issue is not to define any KEY or NUMERIC RANGE fields: make them ordered instead. This has the associated advantage, if you use FILEORG X'100' files, of permitting as many as 32,000 field names in a file.
Number of data sets
You can dynamically add datasets to a Model 204 file.
It is better not to have to do this, because there may be JCL containing file references that would need to be updated at the same time.
Unless space is at a premium, it is a good idea to define larger data set(s) than you need, which gives you the ability to automatically, or manually, increase the Tables without issue.