Indexes are used to sort data logically to improve the speed of searching and sorting operations. The best way to understand indexes is to envision the index at the back of a book (this book, for example).
Suppose you want to find all occurrences of the word datatype in this book. The simple way to do this would be to turn to page 1 and scan every line of every page looking for matches. Although that works, it is obviously not a workable solution. Scanning a few pages of text might be feasible, but scanning an entire book in that manner is not. As the amount of text to be searched increases, so does the time it takes to pinpoint the desired data.
That is why books have indexes. An index is an alphabetical list of words with references to their locations in the book. To search for datatype, you find that word in the index to determine what pages it appears on. Then, you turn to those specific pages to find your matches.
What makes an index work? Simply, it is the fact that it is sorted correctly. The difficulty in finding words in a book is not the amount of content that must be searched; rather, it is the fact that the content is not sorted by word. If the content is sorted like a dictionary, an index is not needed (which is why dictionaries don't have indexes).
Database indexes work in much the same way. Primary key data is always sorted; that's just something the DBMS does for you. Retrieving specific rows by primary key, therefore, is always a fast and efficient operation.
Searching for values in other columns is usually not as efficient, however. For example, what if you want to retrieve all customers who live in a specific state? Because the table is not sorted by state, the DBMS must read every row in the table (starting at the very first row) looking for matches, just as you would have to do if you were trying to find words in a book without using an index.
The solution is to use an index. You may define an index on one or more columns so that the DBMS keeps a sorted list of the contents for its own use. After an index is defined, the DBMS uses it in much the same way as you would use a book index. It searches the sorted index to find the location of any matches and then retrieves those specific rows.
But before you rush off to create dozens of indexes, bear in mind the following:
-
Indexes improve the performance of retrieval operations, but they degrade the performance of data insertion, modification, and deletion. When these operations are executed, the DBMS has to update the index dynamically.
-
Index data can take up lots of storage space.
-
Not all data is suitable for indexing. Data that is not sufficiently unique (State, for example) will not benefit as much from indexing as data that has more possible values (First Name or Last Name, for example).
-
Indexes are used for data filtering and for data sorting. If you frequently sort data in a specific order, that data might be a candidate for indexing.
-
Multiple columns can be defined in an index (for example, State plus City). Such an index will be of use only when data is sorted in State plus City order. (If you want to sort by City, this index would not be of any use.)
There is no hard-and-fast rule as to what should be indexed and when. Most DBMSs provide utilities you can use to determine the effectiveness of indexes, and you should use these regularly.
