Consider a table like
create table example(
tx_id int,
product_id int,
fiscal_yr_id int,
amount decimal
)
According to business rules, the uniqueness constraint is defined as (tx_id,product_id,fiscal_yr_id) ; so naturally they are candidates for the composite primary key.
Also assume that all of the queries would be exact queries i.e. where tx_id=--- and product_id=--- and fiscal_yr_id =-- i.e. no range queries and that the selectivity of the fiscal_yr_id is low i.e. say that it only contains two choices 2006,2007
In cases like these, are there alternatives to having the clustered index pk with (tx_id,product_id,fiscal_yr_id) that might offer better performance?
Thanks
~Rohit