Older IBM i databases are, typically, databases only in name: They lack the relationships between tables, have duplicate information, and so on. Let’s start the cleanup process.
Modernizing the database is, most times, a tortuous and somewhat difficult process. It’s difficult to organize (there are too many variables to take into account—existing programs, external data accesses, queries, and so on), it’s difficult to justify to upper management (as this necessary process is not seen as an efficiency or profit-increasing process, sometimes it’s not easy to get the green light from the higher-ups) and finally, it’s also not easy to implement (there’s always the risk, albeit marginal if things are done correctly, of data loss or corruption). I don’t mean to discourage you, quite the contrary! However, this requires preparation, careful planning and, naturally, common sense. So let’s begin with a few important preparation steps.
Start Preparing to Tidy Up Your Database
By now, you’ve probably realized that not everything is as it should be: The file-naming conventions changed over the years, the data relations and constraints are built and enforced by the program instead of residing in the database, and so on. It’s time to start the careful and meticulous process of planning the tidying up of the database. You’ll need to follow these steps:
Define or refine your naming conventions for the future database: tables, views, constraints, and primary and foreign keys. Keep in mind that, with SQL objects, you’re not constrained by the 10-character limit. You can use up to 128 characters to name tables, views, indexes, and other SQL “things.” Just don’t forget that, if you don’t specify a shorter name, the system will create one for you—trust me, you don’t want that. Being able to assign shorter names, particularly for tables and views, is a great way to keep compatibility with the existing programs. For instance, the INVMST physical file can be converted to an SQL table named Inventory_Master, but you can create an alias for your programs to use. You can use the ALIAS SQL instruction as a way to access a specific member of a physical file in SQL, but it can also be used to “rename” a table to shorter name. That’s why you can’t lose track of the current names: You’ll need to keep them as aliases in the “new” database to maintain, at least for some time, the programs working as they did in the “old” database. In the previous example, the programs “know” INVMST, the old, shorter name of the inventory master physical file, but they won’t recognize Inventory_Master, the new, longer name that resulted from the conversion. I’ll explain this in more depth when I talk about the actual DDS-to-DDL conversion process in the next TechTip.
Here’s what I usually use for the “new” short name of the file, which will be recognizable by the existing RPG programs: PF and LF are for the first two letters of physical and logical files, respectively, followed by up to six characters that characterize the file content. For example, I’d use PFITMMST for an Item Master physical file/table. The logical files/views use the same six letters as the physical file they’re built over, and they have a numeric suffix that starts with 01 and goes up to 99. For example, I’d use LFITMMST01 for the logical file with the most commonly used key for the item master physical file. From here, you can get creative and use, for instance, JF for join files. Just make sure you are consistent and coherent in the naming conventions you use. With this, you can have the best of both worlds: Keep the existing 10-character logic, making it more coherent where and when you can and, on the other hand, use longer, user-friendlier names for your tables and views.
Create or extend your data dictionary. Define standard data types and lengths for “kinds of data,” such as amounts, dates, and coefficients. This will help in the future when you need to create new tables or columns. You should also document the naming standards and abbreviations that you’ll be using in the “new” database. The Modernizing Data Access Roadmap redbook offers a few interesting suggestions, in section 4.1.3, on establishing naming conventions for SQL objects. Actually, the whole book is a must-read!
Once you have all of this ready, you can (finally) start converting the DDS files to DDL. That’ll be the topic of the next TechTip of this series. Until then, feel free to talk about your own experience, naming conventions, and other stuff that can help the community, either in the Comments section below or the usual LinkedIn groups where these TechTips usually end up.