Thread: need SQL logic to find out way's to Implement check sum to validate table data is migrated correctly

Hi 
Need SQL logic/different approach method's  to find out way's 
a) to Implement check sum to validate table data is migrated correctly  

Any guidance

Regards
Bharani SV

Hi

Why do you need checksums? Can you not employ a full outer join? My though behind this is that checksumming is quite costly cpuwise and you have to fiddle with each and every relevent column, be it as part of the checksum string be it as part of the joiner. The joiner would have the advantage that you could, if you wanted, also implement logic to tell you precisely in which columns the difference is encountered.

Kind regards

Thiemo

Am 27.10.2023 um 10:56 schrieb Y_esteembsv-forum:
Hi 
Need SQL logic/different approach method's  to find out way's 
a) to Implement check sum to validate table data is migrated correctly  

Any guidance

Regards
Bharani SV

OP might be referring to migrating from, say, Oracle to Postgresql when oracle_fdw is not available.

On 10/27/23 04:34, Thiemo Kellner wrote:

Hi

Why do you need checksums? Can you not employ a full outer join? My though behind this is that checksumming is quite costly cpuwise and you have to fiddle with each and every relevent column, be it as part of the checksum string be it as part of the joiner. The joiner would have the advantage that you could, if you wanted, also implement logic to tell you precisely in which columns the difference is encountered.

Kind regards

Thiemo

Am 27.10.2023 um 10:56 schrieb Y_esteembsv-forum:
Hi 
Need SQL logic/different approach method's  to find out way's 
a) to Implement check sum to validate table data is migrated correctly  

Any guidance

Regards
Bharani SV

--
Born in Arizona, moved to Babylonia.
On 10/27/23 03:56, Y_esteembsv-forum wrote:
Hi 
Need SQL logic/different approach method's  to find out way's 
a) to Implement check sum to validate table data is migrated correctly 

When migrating from Oracle to PostgreSQL, on both the Oracle side and the Postgresql side, I dumped each table (for the big ones, I split them into views) as a tab-separated value file ordered by the primary key and piped that into md5sum.

Had to play a few games with capitalization in blob/bytea columns

That way, I could verify that each table's data was identical.

--
Born in Arizona, moved to Babylonia.