Though I understand the concepts (which were presented eloquently), I would like to know more about how the table is actually used to dynamically recreate the partitions. Given the fact that you would probably only need to partition a table in the first place if it were quite large, would you actually let some process move your data around to repartition it without monitoring it?

I would also like to see some examples as to how you can dynamically add these partitions, since I plan on using a more predictable partitioning scheme (i.e. adding a new partition per quarter for a date-partitioned table). Can anyone provide such an example? Thanks!