Hi fellow progress talkers,
OpenEdge 10.1B
AIX.Linux
We have had this great idea to make our application able to cope with multiple (financial) administrations. In effect this means that many of the 600+ tables needed an extra field (admincode). Most of these fields are primary key fields.
We have this reconfiguration procedure for new releases which consists of loading df files, loading data in systems tables and post dataprocessing. The problem we are facing now is that all records in the tables with this new admincode field need to have admincode = '01'. What would in your opinion be the fastest way to do this?
In our standard reconfiguration procedure we update the tables in the postdf phase of the reconfiguration procedure with a single user connection on a shutdown database running a program which updates all necessary tables. For relatively small databases of 3-5 GB this process takes some 13 hours to finish. WhichIMO is unacceptable long.
Would ascii dump, change .d file, load data, rebuild indexes be fast?
Would inactivate indexes, update data, rebuild indexes be fast?
Should I start up my database before udating?
Any suggestions would be welcome.
TIA,
Casper.
BTW, I know I could do tests with all possible solutions and I know it is also very much system dependent, but since the process takes so much time to run I am hoping to find some plausible solutions here, which can give me a starting point to start my tests.... (bit lazy )
OpenEdge 10.1B
AIX.Linux
We have had this great idea to make our application able to cope with multiple (financial) administrations. In effect this means that many of the 600+ tables needed an extra field (admincode). Most of these fields are primary key fields.
We have this reconfiguration procedure for new releases which consists of loading df files, loading data in systems tables and post dataprocessing. The problem we are facing now is that all records in the tables with this new admincode field need to have admincode = '01'. What would in your opinion be the fastest way to do this?
In our standard reconfiguration procedure we update the tables in the postdf phase of the reconfiguration procedure with a single user connection on a shutdown database running a program which updates all necessary tables. For relatively small databases of 3-5 GB this process takes some 13 hours to finish. WhichIMO is unacceptable long.
Would ascii dump, change .d file, load data, rebuild indexes be fast?
Would inactivate indexes, update data, rebuild indexes be fast?
Should I start up my database before udating?
Any suggestions would be welcome.
TIA,
Casper.
BTW, I know I could do tests with all possible solutions and I know it is also very much system dependent, but since the process takes so much time to run I am hoping to find some plausible solutions here, which can give me a starting point to start my tests.... (bit lazy )