Several Snapshot-based calls have been implemented with queues. These functions were implemented to increase network performance for unbuffered nodes. The queue calls maintain an in-memory queue of snapshot values, and the values are sent to the home node in one message when the queue is full. This prevents the decrease in performance caused by many smaller network messages.
Typically, applications should be written using the standard snapshot functions. The queuing functions should only be used when performance problems arise and the program is designed to run on an unbuffered node.
For more information, see API Node Buffering.