Indexes are optimized according to the number of unique (distinct) values they include.
FP_NBIT_AUTOSIZE_LIMIT limits the number of distinct values that load as NBit
FP_NBIT_LOOKUP_MB sets a threshold for the total NBit dictionary size
FP_NBIT_ROLLOVER_MAX_MB sets the dictionary size for implicit NBit rollovers from NBit to Flat FP
FP_NBIT_ENFORCE_LIMITS enforces NBit dictionary sizing limits. This option is OFF by default
When the number of distinct values reaches certain levels, choose indexes according to the recommendations in this table.
Number of Unique Values |
Recommended Index Type |
---|---|
Below 1,000 |
LF (HG if table has <100,000 rows) |
1000 and over |
HG and/or HNG |
Columns that hold marital status have just a few unique values (single, married, NULL)
Columns that hold state or province names have fewer than 100 unique values
Columns that hold date data probably have more than 100 but fewer than 65536 unique values
Columns that hold account numbers or social security numbers may have thousands or millions of unique numbers