[BioSQL-l] trouble cramming genbank

[Elia Stupka]

> even more dramatic slowdown.) With the main trunk version of biosql,
> you could also drop all the FK constraints and re-create them after
> loading finished.

I know I will be saying the most unpopular thing on earth and be shot for
not knowing where the righteous path is, however...

...for databases such as biosql and ensembl we have been using so far the
oh so-non-sophisticated and without FK databases and things have been
fine and from what I understand from this mail, things run a lot
faster. Why have things been fine? Because we use biosql as an SRS
replacement, i.e. storing and fetching all sequences, and because ensembl
is a lots-of-work-then-release database, and then it gets used and soon
replaced by something else.

Basically it seems to me we tend to sweat on what to store, and once
things are stored we want to fetch them and do not much else with them.
FKs,etc. and prevention of orphans,etc. are good for general cleanliness,
and they are very good in a database that gets actively updated,
modified,etc. but that doesn't usually happen with biosql, we tend to load
things up, use them and at some point throw them away.

So far in-house we are still on the good old branch I am wondering whether
we should think about keeping a lighter non-FK rich version for
quick-rough access?

Elia

********************************
* http://www.fugu-sg.org/~elia *
* tel:    +65 6874 1467        *
* mobile: +65 9030 7613        *
* fax:    +65 6779 1117        *
********************************