[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Fwd: Re: [Gnumed-devel] Questions About GNUmed DataBase

From: Karsten Hilbert
Subject: Re: Fwd: Re: [Gnumed-devel] Questions About GNUmed DataBase
Date: Mon, 25 Jan 2010 17:27:55 +0100

> > The one thing I did test back in the days (64 MB RAM, PG 7.1) was to
> > search for a patient among about 150k records. That was
> > near instantaneous.
> >
> >
> Performance is only really a problem when joins are performed by the db.
> If
> joins occur on non indexed values then this can really decrease
> performance.
> I've seen joins take >30 secs to occur without indexing and brought down
> to
> sub seconds with indexes.

I know.

> No doubt the searching for patients did not require any joins and was just
> done by querying on maybe first name or last name.

It did but not many. Searching for name parts does require a join
in GNUmed.

> DB performance tuning with large datasets usually always uncovers some
> accidentally missed indexes. Maybe it is a good exercise for Ahmad and his
> uni buddies to undertake :-) They will need some sample queries with joins
> which they can simply find from the code.

I believe that's a great idea !

It would also be useful to benchmark typical clinical queries.


Jetzt kostenlos herunterladen: Internet Explorer 8 und Mozilla Firefox 3.5 -
sicherer, schneller und einfacher!

reply via email to

[Prev in Thread] Current Thread [Next in Thread]