Optimizing Sort Operations - Oracle Database 11gR2 Performance Tuning - page 218

We can see that the SORT UNIQUE and SORT GROUP BY operations are performed in both
situations against the corresponding HASH UNIQUE and HASH GROUP BY . The previous
operations don't return the rows ordered, so a subsequent SORT pass would be required,
using SORT UNIQUE and SORT GROUP BY . The SORT operation is executed to obtain the
unique and group-by functions, so the rows are already ordered.

Here, we can also see that the index IX_CUST_CITY will be used to avoid the full table
scan operation on the table, but it won't be useful in avoiding the sort operation.

Some developers overuse the DISTINCT keyword to ensure
a unique set of results. We can avoid this situation by using
the correct JOINS operations and FOREIGN KEY constraints,
resulting in an optimal execution plan.

See also

F See Chapter 3 , Optimizing Storage Structures for recipes on indexing
F Full Table Scan operation was explained in Chapter 4 , Optimizing SQL Code in the
Avoiding Full Table Scans recipe

Writing top n queries and ranking

One common problem when developing database applications is to show the first n rows of a
set, ordering the data in a specific manner. For example, if we want to see the last 10 articles
submitted in a web application.

In this recipe, we will see how to obtain this scope and how to obtain it faster.

How to do it...

The following steps will demonstrate how to get the top n queries and their ranking:

1. Connect to the SH schema:

CONNECT sh@TESTDB/sh

2. Select the first 10 customers, ordered by their age, from youngest to oldest:
SELECT CUST_ID, CUST_FIRST_NAME, CUST_LAST_NAME,
CUST_YEAR_OF_BIRTH
FROM CUSTOMERS
WHERE ROWNUM < 11
ORDER BY CUST_YEAR_OF_BIRTH DESC;