On Tue, 2004-07-13 at 02:06, Karsten Hilbert wrote:
> Calle Hedberg wrote:
> > The Health (Management) Information Systems Programme I'm working with now
> > covers wholly or partially countries/states with around 200 mill people. We
> > are moving towards DBMS independence for our solutions, but MySQL at the
> > moment seems to be the logical FOSS option (many states/provinces are
> > already using ORACLE, MS SQL or DB2, so it obviously makes sense for them to
> > host health data in the same DBMS).
>
> I am completely baffled as to why this makes MySQL the logical
> FOSS option ?!?

Somewhere on the MySQL web site there is a history of their product,
which reveals that they originally built it purely for decision support
purposes (i.e. reporting and analysis). Indeed the design and feature
set of MySQL up to v3.x reflects this type of work, in which data is
usually collected by other systems, and then bulk-loaded or otherwise
assembled in batches into the decision support database. The emphasis is
on the speed of fairly simple queries. My understanding of the Health
(Management) Information Systems Programme with which Calle is involved
is that it is a system which collects and aggregates summarised and unit
record data collected by clinics, community health centres, hospitals
and other health care facilities into a local or regional decision
support database. Is that correct, Calle? If so, then MySQL is a good
choice.

For systems, such as GnuMed, which are designed to be primary data
collections, then a database with much better and well-proven
transaction processing capabilities, such as PostgreSQL, is a better
choice. MySQL is adding such facilities in V4.x and v5.x, but PostgreSQL
has had them for over a decade, so for now, I know which one I would
prefer to use for the collection of primary, transactional data. But for
"secondary" data, MySQL is fast and reliable. BTW, MySQL seems to have
become popular for the first generation of Web sites because data-driven
Web sites are typically very read-intensive, and MySQL reads data very
quickly. The few updates required in first generation data-driven Web
sites also tended to be fairly simple. But now people want to do much
more complex (primary) data collection through the Web, such as building
EMR/EHRs, which is why MySQL is under pressure to add transactional
processing capabilities. I dare say that it's transactional capabilities
will mature quite quickly, given the size of its user base. Then we'll
have at least two mature transaction-capable FOSS databases. And there
is also Firebird (previously Interbase) and SAPdb (or is that part of
MySQL now). And last time we checked, the query speed of PostgreSQL was
not far behind MySQL on typical decision support queries - the limiting
factor there tends to be disc access speed, not the database, anyway.
-- 

Tim C

PGP/GnuPG Key 1024D/EAF993D0 available from keyservers everywhere
or at http://members.optushome.com.au/tchur/pubkey.asc
Key fingerprint = 8C22 BF76 33BA B3B5 1D5B  EB37 7891 46A9 EAF9 93D0



Reply via email to