Help Center/ GaussDB/ Developer Guide(Centralized_8.x)/ Configuring GUC Parameters/ GUC Parameters/ Query Planning/ Optimizer Cost Constants
Updated on 2024-06-03 GMT+08:00
View PDF
Share

Optimizer Cost Constants

This section describes the optimizer cost constants. The cost variables described here are measured on an arbitrary scale. Only their relative values matter, therefore scaling them all up or down by the same factor will result in no change in the optimizer's choices. By default, they use the cost of fetching sequential pages as the basic unit. That is, seq_page_cost is set to 1.0 and the other cost variables are set with reference to the parameter. However, you can use a different scale, such as actual execution time in milliseconds.

seq_page_cost

Parameter description: Specifies the optimizer's estimated cost of a disk page fetch that is part of a series of sequential fetches.

This is a USERSET parameter. Set it based on instructions provided in Table 1.

Value range: a floating-point number ranging from 0 to DBL_MAX.

Default value: 1

random_page_cost

Parameter description: Specifies the optimizer's estimated cost of an out-of-sequence disk page fetch.

Parameter type: floating point

Unit: none

Value range: 0 to DBL_MAX

Default value: 4

Setting method: This is a USERSET parameter. Set it based on instructions provided in Table 1.

Setting suggestion: none

Although the server allows you to set random_page_cost to a value less than that of seq_page_cost, it is not physically sensitive to do so. However, setting them equal makes sense if the database is entirely cached in RAM, because in that case there is no penalty for fetching pages out of sequence. Also, in a heavily-cached database you should lower both values relative to the CPU parameters, since the cost of fetching a page already in RAM is much smaller than it would normally be.

  • This value can be overwritten for tables and indexes in a particular tablespace by setting the tablespace parameter of the same name.
  • Reducing this value relative to seq_page_cost will cause the system to prefer index scans and raising it will make index scans relatively more expensive. You can increase or decrease both values together to change the disk I/O costs relative to CPU costs.

cpu_tuple_cost

Parameter description: Specifies the optimizer's estimated cost of processing each row during a query.

This is a USERSET parameter. Set it based on instructions provided in Table 1.

Value range: a floating-point number ranging from 0 to DBL_MAX.

Default value: 0.01

cpu_index_tuple_cost

Parameter description: Specifies the optimizer's estimated cost of processing each index entry during an index scan.

This is a USERSET parameter. Set it based on instructions provided in Table 1.

Value range: a floating-point number ranging from 0 to DBL_MAX.

Default value: 0.005

cpu_operator_cost

Parameter description: Specifies the optimizer's estimated cost of processing each operator or function executed during a query.

This is a USERSET parameter. Set it based on instructions provided in Table 1.

Value range: a floating-point number ranging from 0 to DBL_MAX.

Default value: 0.0025

effective_cache_size

Parameter description: Specifies the optimizer's assumption about the effective size of the disk cache that is available to a single query.

When setting this parameter you should consider both GaussDB's shared buffers and the kernel's disk cache. Also, take into account the expected number of concurrent queries on different tables, since they will have to share the available space.

This parameter has no effect on the size of shared memory allocated by GaussDB. It is used only for estimation purposes and does not reserve kernel disk cache. The value is in the unit of disk page. Usually the size of each page is 8192 bytes.

Parameter type: integer.

Unit: page (8 KB)

Value range:

  • Method 1: Set this parameter to an integer without a unit. The integer ranges from 1 to 2147483647. For example, 200 indicates that there are 200 pages, that is, the size of 200 x 8 KB.
  • Method 2: Set this parameter to a value with a unit. The value ranges from 1 x 8 KB to 2147483647 x 8 KB. For example, the value 200MB indicates 200 MB. The unit can only be KB, MB, or GB.

Default value:

280GB (196-core CPU/1536 GB memory); 180GB (128-core CPU/1024 GB memory, 104-core CPU/1024 GB memory, 96-core CPU/1024 GB memory); 135GB (96-core CPU/768 GB memory); 100GB (80-core CPU/640 GB memory); 90GB (64-core CPU/512 GB memory); 80GB (60-core CPU/480 GB memory); 40GB (32-core CPU/256 GB memory); 18GB (16-core CPU/128 GB memory); 8GB (8-core CPU/64 GB memory); 4GB (4-core CPU/32 GB memory); 2GB (4-core CPU/16 GB memory)

Setting method: This is a USERSET parameter. Set it based on instructions provided in Table 1.

Setting suggestion: A value greater than the default one may enable index scan, and a value less than the default one may enable sequential scan.

allocate_mem_cost

Parameter description: Specifies the query optimizer's estimated cost of creating a hash table for memory space using hash join. This parameter is used for optimization when the hash join estimation is inaccurate.

This is a USERSET parameter. Set it based on instructions provided in Table 1.

Value range: a floating-point number ranging from 0 to DBL_MAX.

Default value: 0

Parent topic: Query Planning