Support to create checkpoint for the whole cluster. 
We could use command in console to create/drop/list checkpoints.

To restore from some checkpoint,  users should stop the whole cluster and override data dir with related snapshot dir.  Of course, we will give a tool for that later.