The number of keys can be reduced using

KEY_LEVEL

s for example

KEY_LEVEL=standard,limited

.

As an example consider the case where the index was created using

Organization_Name

and the match uses

Organization_Name

and

Address_Part1

.

The composite key could use

Organization_Name

,

Address_Part1

with

KEY_LEVEL=standard,limited

and then the search could use

KEY_LEVEL=standard SEARCH_LEVEL=Narrow

. This will have a tighter search on

Address_Part1

which will be similar to a Conservative match when the key was built only on

Organization_Name

.

An alternative would be to build the key for

Address_Part1

,

Organization_Name

with

KEY_LEVEL=limited,standard

.

The fewer keys built by the left most fields in the composite keys the smaller the overall number of keys will be. The performance improvement comes from the smaller candidate sets, as the keys will be more selective and less records will need matching. The same (or similar) number of records will be found using composite key with an increase in performance.

Adding extra key / search options using keywords described below will find variations that might be missed (such as for fields with null or empty values -

INSSERTNULLCLOSINGRANGE

and

INSERTNULLCLOSINGKEY

- or variations with initials -

PROBESMAJORINIT1

). Adding these extra key words at search time will retrieve more candidates with minimal impact on performance overall. See examples below in "keywords" section.