1. When a controlled shutdown is executed or before issuing an emergency shutdown (-byF) disconnecting all connected processes that can be disconnected first may help to lighten the load a quicker recovery. The following Article provides a code example of how this may be implemented:
2. During BI recovery, the BI file may grow to as much as twice its current size to complete crash recovery in the UNDO phases.
If the transaction that is handling the crash recovery is stopped, the next crash recovery will undo the pending crash recovery transaction and previous pending transactions, taking more time for crash recovery and more growth of the BI.
3. There are no formulas to calculate how much time will take the crash recovery process, it usually takes as much time as the time the transactions took to the point of the crash and hindered when there are i/o contention on disk. For further advice refer to Article:
