There are two aspects of Liberty InstantOn first response time that need to be improved:
We need to fix the transaction manager related overheads that started appearing after restore when we became functionally correct by running the transaction manager logic that was incorrectly (in prior InstantOn beta releases) not being run on the restore side.
We need to continue to look for further restore time improvements to push the factor of improvement to be higher than 10x (what we targeted with the initial GA). Something closer to 15x may have to be an initial assumption for the target but we may need to revise that as we make further progress with the analysis of the restore side path length.
vijaysun-omr
commented
12 months ago
There are two aspects of Liberty InstantOn first response time that need to be improved:
We need to fix the transaction manager related overheads that started appearing after restore when we became functionally correct by running the transaction manager logic that was incorrectly (in prior InstantOn beta releases) not being run on the restore side.
We need to continue to look for further restore time improvements to push the factor of improvement to be higher than 10x (what we targeted with the initial GA). Something closer to 15x may have to be an initial assumption for the target but we may need to revise that as we make further progress with the analysis of the restore side path length.