We keep a queue in front of the storage engine in order to prevent submitting too many concurrent operations. Going forward, this concept may apply more generally and we employ a similar mechanism elsewhere. Thus, it would make sense to have a general interface for this idea. This interface will likely have some sort of pluggable "policies", for instance for adjusting how operations are dequeued and for adjusting the amount of concurrency. This may even be similar to the current TicketHolder interface, but more generalized.