[Info-Ingres] The little row that wasn't there

[--CELKO--]

I would much prefer table partitioning if I have to keep a lot of live
data which has different hit rates.  This is usually temporal (i.e.
most recent data is also the most used), but can be on other
dimensions, such as geography  (i.e. most of the work comes from a
subset of stores).

Archives are tricky since they aer usually compressed and not easy to
put back into an active data store.  One exception is the SAND data
warehouse products.