HELP: Estimated cost doesn't related to execution time

HELP: Estimated cost doesn't related to execution time

Post by Vladimir Evdokimo » Wed, 05 Apr 2000 04:00:00



Hello,

I test the same query in two similar databases. This query is based on one
table (about 2 millions of records) in the 1st DB, and on the same but
normalized table (with related tables) in the 2nd. These tables have
different indexes.

So when I run that query with "show execution plan" option (result sets are
the same).

use test_one
{quey}

use test_two
{quey}

Query on test_one is faster than on test_two approximately in 1.5 times,
however the execution plan shows that the first query takes 85% of batch
time and the second takes 15%.

What's up?

Thank you,
Vladimir.

 
 
 

HELP: Estimated cost doesn't related to execution time

Post by Jaime McGeath » Tue, 18 Apr 2000 04:00:00


a select against a denormalized table will normally run faster than going
against mutiple table with normalized structure.  That's why in analytical
systems it can be advantageous to denormalize your data.  As far as the
batch time... in test 1 the single table IS the batch, it will be most of
the time. In test two the main table is one part of the batch, the joins to
the other tables will take the most time.  It sounds like your getting
reasonably normal behavior.

Jaime

> Hello,

> I test the same query in two similar databases. This query is based on one
> table (about 2 millions of records) in the 1st DB, and on the same but
> normalized table (with related tables) in the 2nd. These tables have
> different indexes.

> So when I run that query with "show execution plan" option (result sets
are
> the same).

> use test_one
> {quey}

> use test_two
> {quey}

> Query on test_one is faster than on test_two approximately in 1.5 times,
> however the execution plan shows that the first query takes 85% of batch
> time and the second takes 15%.

> What's up?

> Thank you,
> Vladimir.



 
 
 

1. HELP: Estimated cost doesn't related to execution time

Hello,

I test the same query in two similar databases. This query is based on one
table (about 2 millions of records) in the 1st DB, and on the same but
normalized table (with related tables) in the 2nd. These tables have
different indexes.

So when I run that query with "show execution plan" option (result sets are
the same).

use test_one
{quey}

use test_two
{quey}

Query on test_one is faster than on test_two approximately in 1.5 times,
however the execution plan shows that the first query takes 85% of batch
time and the second takes 15%.

What's up?

Thank you,
Vladimir.

2. Linking 3 Tables

3. Understanding the estimate cost fields in the execution plan

4. binary data storage using file index

5. Row Estimate/ cost estimate

6. Help ---- Parallel Database

7. Estimating query execution time

8. Encription - encript.fxp (0/1)

9. estimating the execution time of a query ?

10. Estimated Query Execution TIME

11. Estimating query execution time