Re: The TransRelational Model: Performance Concerns

Alfredo Novoa wrote:
>
> But this is not a realistic assumption and he did not study what
> happens if we have a lot of RAM. [...]

If we assume a lot of RAM then we are essentially talking about a main-memory database in which case you should compare it to the usual techniques for those types of databases. One that springs to my mind is the Monet DB from the CWI in Amsterdam where relations are essentially split in binary relations between tuple identifiers and attribute values. For such relations you can then apply the usual techniques of hashing, indexes et cetera. I strongly suspect that the space and time complexity will be essentially equivalent with the TRM approach, e.g., single attribute equi-joins are also here extremely fast, but it is much less complex and doesn't require fancy allocation algorithms, leaving gaps for inserts, nightly reorganizations or whatever.

In case anyone is interested, they have a web-site:

   http://monetdb.cwi.nl/

There is an open-source implementation on sourceforge if you want to look for yourself, and a boat-load of research literature you can read about how things are done. No patents as far as I can see.

• Jan Hidders