I couldn't find the note you suggested: 837812.1 Is it something like this one? How to recreate the smon_scn_time table so it's not a clustered table to avoid ORA-2475 (Doc ID 2389804.1)
Thank you for the wonderful videos. Looks like that SCN table uses a feature that I have been look for a long time: the ability to retire old records. How is it doing that? I have devised a few mechanisms to purge our old data, but I am not particularly happy about any of those. I would like rows to retire after a date, and if you have this capability internally, why can't make it official and give us a table characteristic like "max record retention 90 days".
These functions are saviours when your table is updated by someone else's code/process without logging anything in the last update time column of your table or any other log tables. Now we can at least tell that some other processes have updated the record.
Great hint, thanks
Glad it was helpful!
I couldn't find the note you suggested: 837812.1
Is it something like this one?
How to recreate the smon_scn_time table so it's not a clustered table to avoid ORA-2475 (Doc ID 2389804.1)
yes
Do you have a video/blog 😊on the row level dependency clause?
No, but you just create a table specifying "rowdependencies". That we track the scn per row not per block. (It costs you a little extra space per row)
Thank you for the wonderful videos. Looks like that SCN table uses a feature that I have been look for a long time: the ability to retire old records. How is it doing that? I have devised a few mechanisms to purge our old data, but I am not particularly happy about any of those. I would like rows to retire after a date, and if you have this capability internally, why can't make it official and give us a table characteristic like "max record retention 90 days".
With rowarchival and ILM you could probably manage this
www.oracle.com/a/ocom/docs/database/database-11g-managing-storage-wp-354099.pdf
These functions are saviours when your table is updated by someone else's code/process without logging anything in the last update time column of your table or any other log tables. Now we can at least tell that some other processes have updated the record.
Definitely!
BTW also blockchain tables
true, but thats a hammer solution to a problem that is not necessarily a nail :-)