By Brian Edminster
I was always jealous of shops that could afford good tools.
Let me explain. Awhile back, I read about HP's history of trying to launch a successor to IMAGE. It was supposed to be called HPIMAGE. It was supposed to be slicker than... well, it was supposed to have all the ability to dynamically index and/or restructure your data that a modern SQL relational database managment system allows, without losing the speed and robustness that makes TurboIMAGE famous. I can recall a few times that having the ability to dynamically restructure a database (while it's in production!) would have been handy. (See: zero downtime)
Then again, a well designed database in a stable application normally shouldn't need that sort of thing with any sort of regularity. Lately, I'm seeing the need to re-structure/alter indexing as a symptom of not knowing your data's demographics and/or designed usage patterns -- especially as the application's data volumes grow.
This need to restructure is also a side-effect of trying to use a single database both as an operational data store (current data only, for day to day production), as well as for research/reporting data warehousing -- where the data is relatively static, but may go back years. Again, that's lazy design. Don't try to make a sports car have the hauling capacity of a truck. You'll end up with neither.
What changes we did need to make, were done with:
1) DBUNLOAD/DBUTIL, PURGE/DBSCHEMA/DBUTIL, CREATE/DBLOAD -- if we were poor (and couldn't afford Adager or other similar tools), or
2) DICTDBU/DBUTIL, PURGE/DBSCHEMA/DBUTIL, CREATE/DICTDBL. This allowed unloading to a tape or disk file -- so if we had enough free space, we could skip using tape, and it was much faster! Also allowed simple re-structuring of the database.
We could do the adding, moving, deleting, and changing the type of datasets; and adding/removing paths, and/or re-arranging order of items in a set. Unfortunately, this was only present if we were lucky enough to be users of Dictionary/3000, or the HP Customizer technology products like MM or HP's Financial software.
3) Best and fastest of all, Model 2 Adager. This even allows transforming the data types, in addition to adding new elements or sets.
But there are still very useful tools that remain on any HP 3000 which still has Predictive Support. Tools you might not know you’ve got.
The Predictive Support files in the SYS account include two very useful tools. While auditing the content of a system, I found :
PSQUAD.PRED.SYS (yep, that's a March 1992 'CM' version of Quad, the customizable editor credited as being developed by Jim Kramer of Quest Systems and Kenneth Stout of Summit Information Systems). It's no QEDIT, to be sure. But Quad sure beats having to use EDIT/3000.
There’s also PSUNLDDB.PRED.SYS and PSLDDB.PRED.SYS. Believe it or not, these are re-named versions of DICTDBU and DICTDBL!
To use these,
- copy PSUNLDDB.PRED.SYS to DICTDBU.PUB.SYS,
- copy PSLDDB.PRED.SYS to DICTDBL.PUB.SYS, and
- copy DICTCAT.PRED.SYS to DICTCAT.PUB.SYS
Okay, perhaps moving these files is bordering on unintended use, and not considered kosher. In that case, set a file equation for the catalog (file dictcat.pred.sys=dictcat.pub.sys), and alter file and group security so you can run the files as they sit.
Either way, this gives you a tool that beats DBUNLOAD/DBLOAD for database capacity maintenance and manipulation — if should you be unfortunate enough to not have the proper tools like Adager, or even DBGeneral.