## Density Calculation on the statistics page.

### Density Calculation on the statistics page.

Hi,
Does anyone know how to calculate the density for an index in SYBASE.

Regds,
R.R

### Density Calculation on the statistics page.

>Hi,
>  Does anyone know how to calculate the density for an index in SYBASE.

>Regds,
> R.R

Density is defined as the nuber of duplicate values in the index.  So to
calculate the density of your index by hand  it is the:
"inverse of the number of unique values"

Take the number calculated above and multiply it by the number of rows in

20 unqiue values = 1/20 =.05 or 5% unique

15,000 rows in table

.05*15000=750 rows on average returned if that index is used.

Josef Richberg
Sybase Professional Services

So if you have 5 unique values you

Scenario:
Created a clustered index and nonclustered indexes on a
table with 5,953,675 rows using a fillfactor of "95" for
clustered index and "98" for nonclustered indexes.
After 82,119 inserts (.01 of table), the nonclustered
indexes had a value of "83" for scan density, "4" for
logical frag, and "96" for avg. page density.  The
clustered index had a value of "96" for scan density, "1"
for logical frag, and "88" for avg. page density.
Methinks I should have used a fillfactor for the
nonclustered indexes a value of "96", and a fillfactor
value of "96" for the clustered index.  The table and
indexes span multiple files.  Confused on trade-off on
scan density versus avg. page density with nightly inserts.
Am in an OLAP (Data Warehouse) environment.  Have other
tables with similiar scenarios (roughly one percent of
table being done nightly with inserts).
Am open to anything.  Thank you