Re: TRM - Morbidity has set in, or not?

"Marshall Spight" <marshall.spight_at_gmail.com> wrote in message news:1147490980.634417.4870_at_i40g2000cwc.googlegroups.com...
> paul c wrote:
> > Marshall Spight wrote:
> > >
> > > The "transrelational" stuff doesn't have much written about it. I
can't
> > > find anything to suggest that it's anything besides a
> > > traditional column store.
> >
> > [...] Still, I've seen a
> > few implementations towards similar ends and at one time followed a lot
> > of the literature and I've never actually seen anybody implement or
> > describe anything quite like it, so even though it seems obvious it
> > doesn't look mainstream traditional to me.
>
> My understanding is that the column store technique dates from perhaps
> the 1970s, and has been used in many special-purpose stores over
> the years. (This is not the same as being part of a dbms, but it's
> still use.)
>
>

Marshall,

I'm going to start a new discussion, rather than divert this one. There are several ideas that come together here,
including some that you have brought up in the past.

First, the idea that the index IS the database, from the article the OP referenced.
Second, the idea of "content based addressing" from some of your old posts. Third, the idea that the index and the table are redundant, but not "harmful redundancy" in the sense that term is usually used to mean. Fourth, hardware associative memories.
Fifth the question of whether a "column store" is any more fundamental than a "row store".
Sixth the question of whether Starkey didn't use the "index IS the database" concept in Interbase.

All of the above has little to do with the TRM, but may merit discussion in its own right. Received on Sat May 13 2006 - 18:56:57 CEST