Thanks Cedar, Jan, and Andy.
Actually the setup is something like this...
There are two remote servers-remoteA and remoteB.
The table of remoteA needs to be sychronized with the
table of remoteB all the time (well, there could be an interval).
remoteB will *publish* every changes and remoteA is *subscribe* to it.
These were my previous solutions:
1. Have a program (using PERL & DBI) in remoteA to connect to remoteB and do the synchronization. >>>>>> I can't
buythis 'coz remoteB has too many *hits*. I just can't afford the cost.
2. Have a trigger in remoteB that will output to a file the result of
every sql or the actually sql. >>>>>> My understanding now is that this will not do it because of a
possibletransaction rollback -- thanks again.
As much as possible I want to do the synchronization
*incrementally* (just deal with the difference between remoteA & remoteB).
But I guess I have to do it the hard way.
Here's my third solution. Please comment on this.
KNOWN FACTORS: ^ poor connection >>> the solution should be intelligent enough to handle such
situation.
3RD SOLUTION: ^ Have a script in remoteB to use pg_dump or sql copy and place it in
the crontab. (say every 5 seconds) ^ Have a script in remoteA that will copy the dump.file from remoteB. Place it
inthe crontab and use *scp* (secure copy) for the copying. After dump.file is acquired, have another script to take
careof it.
What do you think? Any better idea?
Thank you.
Sherwin
