With Adaptive Server 12.5.1 and later, during start-up and HA failover, databases are recovered in parallel by multiple recovery tasks. Database recovery is an I/O-intensive process. The time to recover Adaptive Server with parallel recovery depends on the bandwidth of the underlying I/O subsystem. The I/O subsystem should be able to handle Adaptive Server concurrent I/O requests.
With parallel recovery, multiple tasks recover user databases concurrently. The number of recovery tasks is dependent on the configuration parameter max concurrently recovered db. The default value of 0 indicates that Adaptive Server adopts a self-tuning approach in which it does not make any assumptions on the underlying storage architecture. Statistical I/O sampling methods determine the optimal number of recovery tasks depending on the capabilities of the underlying I/O subsystem. An advisory on the optimal number of recovery tasks is provided. If the configuration value is non-zero, Adaptive Server spawns as many tasks as indicated by the configuration parameter.
During parallel recovery, the system administrator can force serial recovery by reconfiguring the parameter to 1. The active recovery tasks drain out after completing the recovery of the database that is being worked on. The remaining databases are recovered serially.
