Question

My DB is SQL SERVER 2008

I have a large table with 100 million rows and 50 columns.

All the datatype is int.

Then , My query is like ….

Select Count(distinct col5) , Count(distinct col8) , Sum(Col 30) , Sum(Col 49) 
Group by Col1 

Select Count(distinct col5) , Count(distinct col8) , Sum(Col 30) , Sum(Col 49) 
Group by Col1,col2

.....


Select Count(distinct col5) , Count(distinct col8) , Sum(Col 30) , Sum(Col 49) 
Group by Col1 ,Col2,Col3,Col4,Col6,Col7

(about 180 queries ...like above)

But the performance is very bad when I use the count distinct.

So , who can teach me how to improve it please?

and in my case how long might be cost with a best solution ?

very thanks your advices ….

Answer 1

These queries (without WHERE clauses) can hardly be optimized since they need to visit every record to calculate the sums no matter what.

Full table scan and a temporary table to hold the results is a best solution and that’s what you most probably have in your plan.