Considering the possibility of forced database shutdown, I configured it as NOCACHE, but it seems to be causing more overhead than expected.
I'm considering switching to a value like CACHE 1000 or CACHE 5000. If some cached values are lost when forcing DB instances to exit, you can delete the sequence and recreate it later to set the last cache value, START WITH, to the new value.
Isn't this a reasonable approach? Or is there an error in my reasoning?
In context, sequences are increasing at a rate of more than 100 per second, and we speculate that using NOCACHE will result in significant overhead due to frequent commitments