The schema cache simply runs queries against the metadata and then stores the information in plain text files in the
directory you put into the program. If the files in the folder are deleted then the queries are run again and the files are
re-generated. If metadata changes you need to delete those files to get them refreshed. This way, on subsequent times
running the application, instead of running the queries for metadata information it grabs it out of the local text files.

There is a feature I have tried so that updating local cache files is automatic. Each time a new connection is made there
is a check for a value of schema cache version number. If the table is not there then the error is igored. If it is there a
value is returned and it is checked against the cache. If different from what is stored in the cache then it will regenerate
the cache files.

Jason Wharton

--------------------------------------------------------------------------------
--------------------------------------------------------------------------------
Hi List,

This message is probably for Jason but if anyone has input on this
please reply.

I need to understand thoroughly how schema cache works in order to
make some improvements to it.

We have done extensive network monitoring to see how does it impact
the network and it looks like it sends thousands of tiny packages of
about 32 to 64 bytes each in order to populate the schema cache
folder. On a low latency network this is not a problem but when the
latency grows this becomes a serious issue.

Connecting to a database with about 204 tables and about 30 fields
each took 2 hours and 30 minutes over a 0.4 latency internet
connection.

We are interested in investing time into this issue and for that we
need to understand how it works indepth.

Thank you.

Calin Pirtea.
Medisys Australia Pty Ltd.
www.medisys.com.au
calin.pirtea@...