On Monday 25 September 2006 07:21, Andrew Cameron wrote:
> If remote server has not cached a copy it pulls it over WAN.
> If remote server has a cached copy it opens that but whilst doing so checks
> each field is current compared
> To primary server. That seems to be quicker than pulling the whole file
> across.

I see. Problematic design. That will never work to satisfactory performance, 
OR will always be risky regarding data integrity (or maybe even both, 
depending on design)

Some developers still seem to be caught in the paradigm of a paper file where 
you need to transfer the whole lot.

When I access a patient record, typically I only access a small fraction of 
the total infomartion stored for that patient - and only the information that 
gets displayed at any one moment gets queried (and transferred) - sometimes 
even less than that, just an aggregate of the queried data.

I think the current gnumed team even works (or discussed working) on 
look-ahead caching of data (you query and transfer instantly what you need in 
order to display the current active screen, and in the background you query 
and transfer what the program anticipates you may need next while machine and 
network are idle) - but for most systems even this is unnecessary overkill - 
JIT queries limited to what actually gets displayed is usually fast enough.

Horst
_______________________________________________
Gpcg_talk mailing list
[email protected]
http://ozdocit.org/cgi-bin/mailman/listinfo/gpcg_talk

Reply via email to