Re: The TransRelational Model: Performance Concerns

From: Alistair Bayley <alistair_at_abayley.org>
Date: Sat, 13 Nov 2004 22:55:32 GMT
Message-ID: <Ehwld.10067$P7.6574@fe2.news.blueyonder.co.uk>

Alfredo Novoa wrote:
>
> The patent does not prescribe anything about how to physically
> implement the tables.
>
> Binary search does not imply a binary tree.
>
> http://c2.com/cgi/wiki?BinarySearch

I assumed that the tables would be implemented by something that gave O(1) lookup time (amortized perhaps, but still more-or-less constant). And following from this, reconstructing a record would also be a constant time operation (or rather, O(n) where n is the number of columns), as you'd have (say) to fetch 1 block per column from the value table, and 1 block per column from the reconstruction table. Is this more-or-less correct, or have I misunderstood?

> The ring structure is only one of the possible structures we may use.

Can tell me some of the others, please?

Thanks,
Alistair. Received on Sat Nov 13 2004 - 16:55:32 CST