Re: [Gnumed-devel] Approaches to maintain clinical data uptime

[Tim Churches]

Karsten Hilbert wrote:
> On Wed, May 03, 2006 at 08:26:29AM -0700, Jim Busser wrote:
>>>>> Otherwise, a backup may slow down the system and there may be  
>>>>> inconsistency with data, since related tables may be changed  
>>>>> while another is being backed up.
> No such crap is heard of in the vicinity of PostgreSQL.
> Would you even consider accepting any such thing to drive
> your clinical database ?

The large database I described in my previous post (the overall context
is described here: http://www.biomedcentral.com/1471-2458/5/141 )
accepts transactions from nearly 60 EDs (emergency depts) on a 24x7
basis - multiple transactions per patient. As a result, there is not a
minute of any day in which the database is not being written to, and
having to take it offline to do backups could cause problems (because
HL7 transaction messages would then queue up in each ED, and may fill up
the HL7 message buffers on local ED systems, resulting in us missing
some data - although there is additional buffering of HL7 messages
elsewhere in the system, and we are only receiving copies of data, so
there is no risk of original data loss and any gaps can always be filled
in post hoc, albeit at some effort). That's one of the reasons why we
chose PG - so the backups could run while the DB was still online.

Tim C