3 ms·
At my previous job (a .edu), our "enterprise" systems still ran on the trusty, decades old mainframe. Batch jobs typically ran once daily and were written in JC
by jlgaddis 14y ago
At my previous job (a .edu), our "enterprise" systems still ran on the trusty, decades old mainframe. Batch jobs typically ran once daily and were written in JCL and getting data out required writing reports in RPG.
Until ~5 years ago, we had a staff member whose job duties involved nothing but using RJE to submit these "report jobs", wait for them to finish, print them off on the huge IBM printers, and deliver them to the appropriate people (who, the majority of the time, never even looked at them).
Coincidentally, it was our marketing department who came up with ideas they wanted to pursue but were unable to due to a "lack of data". The data they needed was actually there but getting it in a form they could work with was the issue. My boss, a man who, it seemed, was incapable of saying "no" to requests, told them we'd come up with something. Thus, the idea for the "Mother Of All Databases" (MOADB) was born.
Fast forward a few months and several meetings with other departments and we had a vision of what this database would look like. My boss, despite having never written a SQL query in his life, had decided to take it upon himself to "design the database". Because I knew the most about databases (which wasn't saying much), I became intimately involved in this project (against my better judgment and recommendations that we ditch the project).
What we ended up with was a database shared amongst tens of users across several different departments. This database consisted of, ultimately, dozens of tables (some stored in Access, some in MS SQL Server), and an MS Access front-end. The boss designed "forms" in Access for the end users to interface with. Remember how I said my boss had never written a SQL query in his life? Well, he also had no idea what normalization is. Several of the tables in the database had in excess of 100 columns (the record, last I remember checking, was 173 columns).
Because I had looked at a report written in RPG once, I was drafted to write a massive report that would pull all of this data out of our mainframe system. The resulting "data files" were CSV files that we had to get into the database. We hit some limit of the system and had to split the job up so that it would only run on a semester's worth of data at a time. The aforementioned staff member would run these jobs every morning, wait for them to finish, and download the data files.
To get the data into the database, CGI scripts were written. The staff member would navigate to a specific URL where he'd upload the CSV data files to the CGI scripts, which would parse the CSV files, and insert/replace the data in the database. An e-mail would then go out to the users of the database, letting them know that the "daily updates" had ran.
This database was ultimately used by many different departments for a variety of tasks: marketing, retention, financial aid, advising, testing, etc. It grew and grew and results to add new data to the database continued to come in (you know how I said my boss didn't like to say "no" to anyone? eventually this changed.). It contained records on tens of thousands of students (including lots of data that shouldn't be laying around in .mdb files on file servers).
Although it was being "phased out" as we had finally upgraded from our old mainframe-based system, the database still existed and was being used whenever I left about two years ago.
(Related note: we dealt with, on an almost daily basis, requests from users for "Excel databases". This comic really strikes a nerve with me!)
Edit: I forgot a step... Staff member would download data (CSV) files, open them in Excel, had to do something to them (can't remember what), save them as .xls files, and upload them to the CGI scripts (Perl, using Spreadsheet::ParseExcel, if memory serves).