The Freelists Parameter and DML Performance Oracle Tips by Burleson Consulting

The freelists parameter tells Oracle how many segment header blocks to create for a table or index. Multiple freelists are used to prevent segment header contention when several tasks compete to insert, update, or delete from the table. The freelists parameter should be set to the maximum number of concurrent update operation.

TIP - Prior to Oracle8i, you must reorganize the table to change the freelists storage parameter. In Oracle8i, you can dynamically add freelists to any table or index with the alter table command, as long as the compatibility parameter is set to at least 8.1.6. In Oracle8i, adding a freelist reserves a new block in the table to hold the control structures.

The freelist groups Storage Parameter for OPS

The freelist groups parameter is used in Oracle Parallel Server (renamed Real Application Clusters in Oracle9i). When multiple instances access a table, separate freelist groups are allocated in the segment header. The freelist groups parameter should be set the number of instances that access the table.

Note: The variables are called pctfree and pctused in the create table and alter table syntax, but they are called PCT_FREE and PCT_USED in the Remote DBA_tables view in the Oracle dictionary. The programmer responsible for this mix-up was promoted to senior vice president in recognition of his contribution to the complexity of the Oracle software.

Summary of Storage Parameter Rules

The following rules govern the settings for the storage parameters freelists, freelist groups, pctfree, and pctused. As you know, the value of pctused and pctfree can easily be changed at any time with the alter table command, and the observant Remote DBA should be able to develop a methodology for deciding the optimal settings for these parameters. For now, accept these rules, and I will discuss them in detail later in this chapter.

There is a direct trade-off between effective space utilization and high performance, and the table storage parameters control this trade-off:

• For efficient space reuse   A high value for pctused will effectively reuse space on data blocks, but at the expense of additional I/O. A high pctused means that relatively full blocks are placed on the freelist. Hence, these blocks will be able to accept only a few rows before becoming full again, leading to more I/O.

• For high performance   A low value for pctused means that Oracle will not place a data block onto the freelist until it is nearly empty. The block will be able to accept many rows until it becomes full, thereby reducing I/O at insert time. Remember that it is always faster for Oracle to extend into new blocks than to reuse existing blocks. It takes fewer resources for Oracle to extend a table than to manage freelists.

While we will go into the justification for these rules later in this chapter, let’s review the general guidelines for setting of object storage parameters:

• Always set pctused to allow enough room to accept a new row. We never want to have free blocks that do not have enough room to accept a row. If we do, this will cause a slowdown, since Oracle will attempt to read five “dead” free blocks before extending the table to get an empty block.

• The presence of chained rows in a table means that pctfree is too low or that db_block_size is too small. In most cases within Oracle, RAW and LONG RAW columns make huge rows that exceed the maximum block size for Oracle, making chained rows unavoidable.

• If a table has multiple simultaneous insert SQL processes, it needs to have multiple simultaneous delete processes.

Running a single purge job will place all of the free blocks on only one freelist, and none of the other freelists will contain any free blocks from the purge.

• The freelist parameter should be set to the high-water mark of updates to a table. For example, if the customer table has up to 20 end users performing insert operations at any time, the customer table should have freelists=20.

• The freelist groups parameter should be set the number of Oracle Parallel Server instances that access the table.

 

This is an excerpt from "Oracle High-Performance SQL Tuning" by Donald K. Burleson, published by Oracle Press.