lee_shelton
New Member
I couldn't find the 'search' button so I'm not sure if this is the right forum...
We want to start replication. We have 500 offices across the country with separate Progress 8.3 (SCO Unix) databases. So we want to 'replicate' those db's by combining them into one large db at the central office.
The path chosen is to install db triggers and have the trigger programs append the updated buffer info (creates/upodates/deletes) to a text file. One text-file per each of the 500 offices.
Then send that text-file over a "dial-up" connection (using maybe FTP) to the central office.
At that point they want to load the text-file into a SQL-Server db, but what has me puzzled is back at the start:
Has anyone used just a "text file" for their replication log trail?
I'm afraid of OS file collisions when multiple users start firing those triggers -- all going to one text file. (Although someone mentioned to me SCO's file-locking functions scripts or C-functions -- I've not looked into it, yet)
Shouldn't we start like the sample code in /usr/dlc with another database and table with the RAW data field -- and then _under Progress control_ export the latest transactions to a text-file??
I don't know how to articulate my concerns and the other team members seem "bent" on NOT distributing another 500 db's (however small).
Thanks
Lee
We want to start replication. We have 500 offices across the country with separate Progress 8.3 (SCO Unix) databases. So we want to 'replicate' those db's by combining them into one large db at the central office.
The path chosen is to install db triggers and have the trigger programs append the updated buffer info (creates/upodates/deletes) to a text file. One text-file per each of the 500 offices.
Then send that text-file over a "dial-up" connection (using maybe FTP) to the central office.
At that point they want to load the text-file into a SQL-Server db, but what has me puzzled is back at the start:
Has anyone used just a "text file" for their replication log trail?
I'm afraid of OS file collisions when multiple users start firing those triggers -- all going to one text file. (Although someone mentioned to me SCO's file-locking functions scripts or C-functions -- I've not looked into it, yet)
Shouldn't we start like the sample code in /usr/dlc with another database and table with the RAW data field -- and then _under Progress control_ export the latest transactions to a text-file??
I don't know how to articulate my concerns and the other team members seem "bent" on NOT distributing another 500 db's (however small).
Thanks
Lee