SQL Indexing Issues

Question

As far as I understand:

The clustered index organizes the data physically by index, so if you use the last name as a cluster index, when you make a choice *, you will get the last names in alphabetical order.

A non-clustered index does not physically reorder your database, but creates some sort of lookup table sorted by the columns you select.

My book says that you can have 16 columns for a clustered index. I would think that you can only select one column, since it physically reorders your database to them? Or multiple columns if the first column contains duplicates?

Isn’t it always faster to use non-clustered indexes, since SQL does not need to shuffle the data around?

Answer 1

The clustered index organizes the data physically by index, so if you use the last name as a cluster index, when you make a choice *, you will get the last names in alphabetical order.

This is not necessarily true. An order is not guaranteed if you have not provided ORDER BYfor your request.

My book says that you can have 16 columns for a clustered index. I would think that you can only select one column, since it physically reorders your database to them? Or multiple columns if the first column contains duplicates?

Sorts lexicographically: first in the first column, then on the second (in the case of binding to the first column, etc.).

, SQL ?

INSERT, , DML (, ..).

, .

Answer 2

- , SQL Server 2005 - , . , "" . , , , , , , .

Answer 3

, , , . , .

, .

, , , , , (, ).