david_dick at iprimus.com.au
Wed Mar 24 23:39:17 CST 2004
I've reached some conclusions about custom reports for a perl web based
system that i think are ok, but i was just wondering if anyone can see
flaws with this approach or has a better one.
Taking for granted the possibility that what i'm working on develops a
user base, i'm assuming that sooner or later, a sophisticated user will
want better reports from the system.
To me, better reporting means either me working 9-5 making reports ("can
you make that red slightly more blue?"), or suits using cognos, ms
access/excel, etc to interrogate the database. Now the technical dramas
of making the connection seem quite small. Build ODBC connectivity into
my database, use ssh port forwarding to establish a secure connection,
blah, blah. The problem i see is this advanced user accessing the
database in ways i had not anticipated, causing whole db to die under
table scans of massive tables and every other user to sit around waiting
for the system to come back.
Possible Solution: The answer seems to be replication in some form, so
i thought the following would be ok. Each night, using $dbh->tables()
from DBI, get a list of all the tables, convert them into csv files
using Text::CSV_XS, zip them up using Archive::Zip and then wait for the
user to request the zip file during the day. The user unzips the
archive in a directory and points the reporting tool at the csv files.
Does this seem ok?
More information about the Melbourne-pm