|Oracle® Database Performance Tuning Guide
10g Release 1 (10.1)
Part Number B10752-01
This chapter provides an overview of data access methods using indexes and clusters that can enhance or degrade performance.
The chapter contains the following sections:
This section describes the following:
Although query optimization helps avoid the use of nonselective indexes within query execution, the SQL engine must continue to maintain all indexes defined against a table, regardless of whether they are used. Index maintenance can present a significant CPU and I/O resource demand in any write-intensive application. In other words, do not build indexes unless necessary.
To maintain optimal performance, drop indexes that an application is not using. You can find indexes that are not being used by using the
USAGE functionality over a period of time that is representative of your workload. This monitoring feature records whether or not an index has been used. If you find that an index has not been used, then drop it. Make sure you are monitoring a representative workload to avoid dropping an index which is used, but not by the workload you sampled.
Also, indexes within an application sometimes have uses that are not immediately apparent from a survey of statement execution plans. An example of this is a foreign key index on a parent table, which prevents share locks from being taken out on a child table.
If you are deciding whether to create new indexes to tune statements, then you can also use the
PLAN statement to determine whether the optimizer will choose to use these indexes when the application is run. If you create new indexes to tune a statement that is currently parsed, then Oracle invalidates the statement.
When the statement is next parsed, the optimizer automatically chooses a new execution plan that could potentially use the new index. If you create new indexes on a remote database to tune a distributed statement, then the optimizer considers these indexes when the statement is next parsed.
Note that creating an index to tune one statement can affect the optimizer's choice of execution plans for other statements. For example, if you create an index to be used by one statement, then the optimizer can choose to use that index for other statements in the application as well. For this reason, reexamine the application's performance and execution plans, and rerun the SQL trace facility after you have tuned those statements that you initially identified for tuning.
The SQLAccess Advisor is an alternative to manually determining which indexes are required. This advisor recommends a set of indexes when invoked from Advisor Central in Oracle Enterprise Manager or run through the
DBMS_ADVISOR package APIs. The SQLAccess Advisor either recommends using a workload or it generates a hypothetical workload for a specified schema. Various workload sources are available, such as the current contents of the SQL Cache, a user defined set of SQL statements, or a SQL Tuning set. Given a workload, the SQLAccess Advisor generates a set of recommendations from which you can select the indexes that are to be implemented. An implementation script is provided that can be executed manually or automatically through Oracle Enterprise Manager.
Oracle Data Warehousing Guide for information on the SQLAccess Advisor
A key is a column or expression on which you can build an index. Follow these guidelines for choosing keys to index:
Oracle automatically creates indexes, or uses existing indexes, on the keys and expressions of unique and primary keys that you define with integrity constraints.
Indexing low selectivity columns can be helpful if the data distribution is skewed so that one or two values occur much less often than other values.
UPDATEstatements that modify indexed columns and
DELETEstatements that modify indexed tables take longer than if there were no index. Such SQL statements must modify data in indexes as well as data in tables. They also generate additional undo and redo.
WHEREclauses with functions or operators. A
WHEREclause that uses a function, other than
MAX, or an operator with an indexed key does not make available the access path that uses the index except with function-based indexes.
DELETEstatements access the parent and child tables. Such an index allows
DELETEs on the parent table without share locking the child table.
DELETEs and the use of the space required to store the index. You might want to experiment by comparing the processing times of the SQL statements with and without indexes. You can measure processing time with the SQL trace facility.
Oracle Database Application Developer's Guide - Fundamentals for more information on the effects of foreign keys on locking
A composite index contains more than one key column. Composite indexes can provide additional advantages over single-column indexes:
Sometimes two or more columns or expressions, each with poor selectivity, can be combined to form a composite index with higher selectivity.
If all columns selected by a query are in a composite index, then Oracle can return these values from the index without accessing the table.
A SQL statement can use an access path involving a composite index if the statement contains constructs that use a leading portion of the index.
This is no longer the case with index skip scans. See "Index Skip Scans".
A leading portion of an index is a set of one or more columns that were specified first and consecutively in the list of columns in the
INDEX statement that created the index. Consider this
xyzcombinations of columns are leading portions of the index
zcombinations of columns are not leading portions of the index
Follow these guidelines for choosing keys for composite indexes:
WHEREclause conditions combined with
ANDoperators, especially if their combined selectivity is better than the selectivity of either key individually.
Of course, consider the guidelines associated with the general performance advantages and trade-offs of indexes described in the previous sections.
Follow these guidelines for ordering keys in composite indexes:
WHEREclauses make up a leading portion.
WHEREclauses more frequently, then be sure to create the index so that the more frequently selected keys make up a leading portion to allow the statements that use only these keys to use the index.
WHEREclauses equally often but the data is physically ordered on one of the keys, then place that key first in the composite index.
Even after you create an index, the optimizer cannot use an access path that uses the index simply because the index exists. The optimizer can choose such an access path for a SQL statement only if it contains a construct that makes the access path available. To allow the query optimizer the option of using an index access path, ensure that the statement contains a construct that makes such an access path available.
In some cases, you might want to prevent a SQL statement from using an access path that uses an existing index. You might want to do this if you know that the index is not very selective and that a full table scan would be more efficient. If the statement contains a construct that makes such an index access path available, then you can force the optimizer to use a full table scan through one of the following methods:
NO_INDEXhint to give the query optimizer maximum flexibility while disallowing the use of a certain index.
FULLhint to force the optimizer to choose a full table scan instead of an index scan.
INDEX_COMBINEhints to force the optimizer to use one index or a set of listed indexes instead of another.
Chapter 17, "Optimizer Hints" for more information on the
Parallel execution uses indexes effectively. It does not perform parallel index range scans, but it does perform parallel index lookups for parallel nested loop join execution. If an index is very selective (there are few rows for each index entry), then it might be better to use sequential index lookup rather than parallel table scan.
You might want to re-create an index to compact it and minimize fragmented space, or to change the index's storage characteristics. When creating a new index that is a subset of an existing index or when rebuilding an existing index with new storage characteristics, Oracle might use the existing index instead of the base table to improve the performance of the index build.
To avoid calling
However, there are cases where it can be beneficial to use the base table instead of the existing index. Consider an index on a table on which a lot of DML has been performed. Because of the DML, the size of the index can increase to the point where each block is only 50% full, or even less. If the index refers to most of the columns in the table, then the index could actually be larger than the table. In this case, it is faster to use the base table rather than the index to re-create the index.
REBUILD statement to reorganize or compact an existing index or to change its storage characteristics. The
REBUILD statement uses the existing index as the basis for the new one. All index storage statements are supported, such as
STORAGE (for extent allocation),
TABLESPACE (to move the index to a new tablespace), and
INITRANS (to change the initial number of entries).
REBUILD is faster than dropping and re-creating an index, because this statement uses the fast full scan feature. It reads all the index blocks using multiblock I/O, then discards the branch blocks. A further advantage of this approach is that the old index is still available for queries while the rebuild is in progress.
Oracle Database SQL Reference for more information about the
You can coalesce leaf blocks of an index by using the
INDEX statement with the
COALESCE option. This option lets you combine leaf levels of an index to free blocks for reuse. You can also rebuild the index online.
You can use an existing nonunique index on a table to enforce uniqueness, either for
UNIQUE constraints or the unique aspect of a
KEY constraint. The advantage of this approach is that the index remains available and valid when the constraint is disabled. Therefore, enabling a disabled
KEY constraint does not require rebuilding the unique index associated with the constraint. This can yield significant time savings on enable operations for large tables.
Using a nonunique index to enforce uniqueness also lets you eliminate redundant indexes. You do not need a unique index on a primary key column if that column already is included as the prefix of a composite index. You can use the existing index to enable and enforce the constraint. You also save significant space by not duplicating the index. However, if the existing index is partitioned, then the partitioning key of the index must also be a subset of the
UNIQUE key; otherwise, Oracle creates an additional unique index to enforce the constraint.
An enabled novalidated constraint behaves similarly to an enabled validated constraint for new data. Placing a constraint in the enabled novalidated state signifies that any new data entered into the table must conform to the constraint. Existing data is not checked. By placing a constraint in the enabled novalidated state, you enable the constraint without locking the table.
If you change a constraint from disabled to enabled, then the table must be locked. No new DML, queries, or DDL can occur, because there is no mechanism to ensure that operations on the table conform to the constraint during the enable operation. The enabled novalidated state prevents operations violating the constraint from being performed on the table.
An enabled novalidated constraint can be validated with a parallel, consistent-read query of the table to determine whether any data violates the constraint. No locking is performed, and the enable operation does not block readers or writers to the table. In addition, enabled novalidated constraints can be validated in parallel: Multiple constraints can be validated at the same time and each constraint's validity check can be determined using parallel query.
Use the following approach to create tables with constraints and indexes:
NULLconstraints can be unnamed and should be created enabled and validated. All other constraints (
KEY) should be named and created disabled.
TABLEstatement for each constraint, validate all constraints. Do this to primary keys before foreign keys. For example,
CREATE TABLE t (a NUMBER CONSTRAINT apk PRIMARY KEY DISABLE, b NUMBER NOT NULL); CREATE TABLE x (c NUMBER CONSTRAINT afk REFERENCES t DISABLE);
Now you can use Import or Fast Loader to load data into table
CREATE UNIQUE INDEX tai ON t (a); CREATE INDEX tci ON x (c); ALTER TABLE t MODIFY CONSTRAINT apk ENABLE NOVALIDATE; ALTER TABLE x MODIFY CONSTRAINT afk ENABLE NOVALIDATE;
At this point, users can start performing
SELECTs on table
Now the constraints are enabled and validated.
Oracle Database Concepts for a complete discussion of integrity constraints
A function-based index includes columns that are either transformed by a function, such as the
UPPER function, or included in an expression, such as
col2. With a function-based index, you can store computation-intensive expressions in the index.
Defining a function-based index on the transformed column or expression allows that data to be returned using the index when that function or expression is used in a
WHERE clause or an
BY clause. This allows Oracle to bypass computing the value of the expression when processing
DELETE statements. Therefore, a function-based index can be beneficial when frequently-executed SQL statements include transformed columns, or columns in expressions, in a
Oracle treats descending indexes as function-based indexes. The columns marked
DESC are sorted in descending order.
For example, function-based indexes defined with the
column_name) keywords allow case-insensitive searches. The index created in the following statement:
facilitates processing queries such as:
Similar to partitioned tables, partitioned indexes improve manageability, availability, performance, and scalability. They can either be partitioned independently (global indexes) or automatically linked to a table's partitioning method (local indexes).
Oracle supports both range and hash partitioned global indexes. In a range partitioned global index, each index partition contains values defined by a partition bound. In a hash partitioned global index, each partition contains values determined by the Oracle hash function.
The hash method can improve performance of indexes where a small number leaf blocks in the index have high contention in multiuser OLTP environment. In some OLTP applications, index insertions happen only at the right edge of the index. This could happen when the index is defined on monotonically increasing columns. In such situations right edge of the index becomes a hotspot because of contention for index pages, buffers, latches for update, and additional index maintenance activity, which results in performance degradation.
With hash partitioned global indexes index entries are hashed to different partitions based on partitioning key and the number of partitions. This spreads out contention over number of defined partitions, resulting in increased throughput. Hash-partitioned global indexes would benefit TPC-H refresh functions that are executed as massive PDMLs into huge fact tables because contention for buffer latches would be spread out over multiple partitions.
With hash partitioning, an index entry will be mapped to a particular index partition based on the hash value generated by Oracle. The syntax to create hash-partitioned global index is very similar to hash-partitioned table. Queries involving equality and
IN predicates on index partitioning key can efficiently use global hash partitioned index to answer queries quickly.
An index-organized table differs from an ordinary table in that the data for the table is held in its associated index. Changes to 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 for queries that involve exact match or range search or both.
Global hash-partitioned indexes are supported for index-organized tables and can provide performance benefits in a multiuser OLTP environment.
Bitmap indexes can substantially improve performance of queries that have all of the following characteristics:
WHEREclause contains multiple predicates on low- or medium-cardinality columns.
You can use multiple bitmap indexes to evaluate the conditions on a single table. Bitmap indexes are thus highly advantageous for complex ad hoc queries that contain lengthy
WHERE clauses. Bitmap indexes can also provide optimal performance for aggregate queries and for optimizing joins in star schemas.
In addition to a bitmap index on a single table, you can create a bitmap join index, which is a bitmap index for the join of two or more tables. A bitmap join index is a space-saving way to reduce the volume of data that must be joined, by performing restrictions in advance. For each value in a column of a table, a bitmap join index stores the rowids of corresponding rows in another table. In a data warehousing environment, the join condition is an equi-inner join between the primary key column(s) of the dimension tables and the foreign key column(s) in the fact table.
Bitmap join indexes are much more efficient in storage than materialized join views, an alternative for materializing joins in advance. This is because the materialized join views do not compress the rowids of the fact tables.
Oracle Data Warehousing Guide for examples and restrictions of bitmap join indexes
Domain indexes are built using the indexing logic supplied by a user-defined indextype. An indextype provides an efficient mechanism to access data that satisfy certain operator predicates. Typically, the user-defined indextype is part of an Oracle option, like the Spatial option. For example, the
SpatialIndextype allows efficient search and retrieval of spatial data that overlap a given bounding box.
The cartridge determines the parameters you can specify in creating and maintaining the domain index. Similarly, the performance and storage characteristics of the domain index are presented in the specific cartridge documentation.
Refer to the appropriate cartridge documentation for information such as the following:
Oracle Spatial User's Guide and Reference for information about the
Clusters are groups of one or more tables that are physically stored together because they share common columns and usually are used together. Because related rows are physically stored together, disk access time improves.
To create a cluster, use the
Oracle Database Concepts for more information on clusters
Follow these guidelines when deciding whether to cluster tables:
Consider the benefits and drawbacks of clusters with respect to the needs of the application. For example, you might decide that the performance gain for join statements outweighs the performance loss for statements that modify cluster key values. You might want to experiment and compare processing times with the tables both clustered and stored separately.
Oracle Database Administrator's Guide for more information on creating clusters
Hash clusters group table data by applying a hash function to each row's cluster key value. All rows with the same cluster key value are stored together on disk. Consider the benefits and drawbacks of hash clusters with respect to the needs of the application. You might want to experiment and compare processing times with a particular table as it is stored in a hash cluster, and as it is stored alone with an index.
Follow these guidelines for choosing when to use hash clusters:
WHEREclauses, if the
WHEREclauses contain equality conditions that use the same column or combination of columns. Designate this column or combination of columns as the cluster key.
Storing a single table in a hash cluster can be useful, regardless of whether the table is joined frequently with other tables, as long as hashing is appropriate for the table based on the considerations in this list.