Re: Generating fake databases

On Oct 15, 1:51 am, Erwin <e.sm..._at_myonline.be> wrote:
> On 14 okt, 12:11, Roy Hann <specia..._at_processed.almost.meat> wrote:
>
>
>
> > I'm not looking for recipes.  I am looking for research that goes beyond
> > naive recipes; that's why I posted to a theory group.
>
> > --
> > Roy
>
> Like Bob, I am not aware of any paperwork on the subject.
>
> When I built my generic testDB construction tool (for the IDMS system,
> not an SQL one), I just had to tackle the problems one by one as they
> came about.
>
> So all in all, all I'll have to offer are the ideas and concepts
> underpinning what I built at the time.  A lot of those do carry over
> to SQL systems, but they still remain little more than "recipes" for
> an actual solution.
>
> Sorry.

On paper, I imagine a completely unrestricted algebra would need to be involved, eg., full support for negation and union, unlike SQL, and the constraints would need to be expressed algebraically, but I doubt there is much demand for such.

Once I had to deal with an add-on db that could interface with other dbs, such as IMS. The problem was to fit the system on a single mainframe instead of the four it needed as written. This was at a very large government installation of an ancient constitutional monarchy. The system recorded various stuff about broken families. There was all kinds of security involved because every 'case' was considered confidential to the families and the case worker assigned to them, plus extra steps were taken for prominent families, with labels such as 'locally secure' and 'nationally secure', eg., several heirs to the throne had divorced parents. I needed to get some kind of profile of the IMS db, volumes by segment type and so forth, and the non-union bureaucrats told me they had a test db that matched the volumes of the production one but with mangled data. They asked the unionized operations people to make a mainframe available to me for sampling purposes. After a couple of days I was kicked off because they needed the cpu back. I thanked one of the operations guys for their trouble and said I hoped it wouldn't take long to restore the production data. He laughed, said it wasn't much trouble because they didn't bother to set up the mangled data, I had been using a production copy all along. Any complaint by the senior civil servants would have been stymied by threats of job action. Received on Sat Oct 15 2011 - 16:26:56 CEST