14.4 Using Index-Organized Tables for Performance

原文：

An index-organized table differs from an ordinarytable in that the data for the table is held in its associated index. Changesto the table data, such as adding new rows, updating rows, or deleting rows,result only in updating the index. Because data rows are stored in the index,index-organized tables provide faster key-based access to table data forqueries that involve exact match or range search or both.

A parent/child relationship is an example of asituation that may warrant an index-organized table. For example, a memberstable has a child table containing phone numbers. Phone numbers for a memberare changed and added over time. In a heap-organized table, rows are insertedin data blocks where they fit. However, when you query the members table, youalways retrieve the phone numbers from the child table. To make the retrievalmore efficient, you can store the phone numbers in an index-organized table sothat phone records for a given member are inserted near each other in the datablocks.

In some circumstances, an index-organized table mayprovide a performance advantage over a heap-organized table. For example, if aquery requires fewer blocks in the cache, then the database uses the buffercache more efficiently. If fewer distinct blocks are needed for a query, then asingle physical I/O may retrieve all necessary data, requiring a smaller amountof I/O for each query.

Global hash-partitioned indexes are supported forindex-organized tables and can provide performance benefits in a multiuser OLTPenvironment. Index-organized tables are useful when you must store relatedpieces of data together or physically store data in a specific order.

譯文：

索引組織的表與普通表的不同之處在於表的資料儲存在相關的索引中。對錶中資料的修改，比如增加新的行，更新已有行，或者刪除行這些操作終究都只是更新索引的操作。因為資料行被儲存在索引中，索引組織的表在查詢時能提供更快的基於key的資料訪問，無論是精確匹配的查詢還是範圍查詢皆是如此。

一種父子關係的情況可能就需要索引組織的表。比如，一個成員表有一個子表，子表中包含成員的電話號碼。隨著時間的推移，成員的電話號碼會改變和增加新的號碼。在一個堆組織的表中，資料行被新增在它們可以被放進去的資料塊中。但是，當查詢成員表時，總是要從子表中獲得電話號碼。為了是獲得電話號碼的操作更加有效，可以將電話號碼儲存在索引組織的表中，這樣一個特定成員的若干個電話號碼會被儲存在相鄰的資料塊中。

在某些情況下，索引組織的表會比堆組織的表提供更好的效能。比如，一個查詢如果需要的快取塊少一些，那麼資料塊就能夠更有效的利用緩衝塊。如果一個查詢需要的資料塊少一些，可能一次I/O就能或得到所有需要的資料，這樣會減少查詢需要的I/O次數。

索引組織的表也支援全域性雜湊分割槽索引，這可以在多使用者的OLTP環境中提供效能優勢。當你必須把相關的資料都儲存在一起或者需要資料物理上有序儲存的時候，你可能需要索引組織的表。