Re: How to handle secondary keys in partitioning?

[Mattias Jonsson]

I'm not sure if I understand what you ask about, but I'll try to explain what I think you are looking for:

Partitioning handles indexes automatically, also secondary indexes, so you do not have to do queries for each partition separately. Partitioning also make use of indexes for lookups and sorting, just like non partitioned tables.

An ordered index read from a partitioned tables is implemented by the partitioning handler ha_partition by:

1) do an ordered index read from each partition that is not pruned away
2) use a priority queue (merge sort) to return the 'first' row
3) do an index next read from the partition which the returned row came from (i.e. keep the priority queue filled with one row per partition).

That way you (as a user) do not have to query each partition your self or sort the result without making use of the index.