[ORACLE] Database Initialization Parameter Sizing recommendations

This section provides sizing recommendations based on the active Oracle E-Business Suite user counts. The following table should be used to size the relevant parameters:
Parameter NameDevelopment or Test Instance11-100 Users101-500 Users501-1000 Users1001-2000 Users
processes20020080012002500
sessions400400160024005000
sga_target Footnote 11G1G2G3G14G
shared_pool_size (csp)N/AN/AN/A1800M3000M
shared_pool_reserved_size (csp)N/AN/AN/A180M300M
shared_pool_size (no csp)400M600M800M1000M2000M
shared_pool_reserved_size (no csp)40M60M80M100M100M
pga_aggregate_target1G2G4G10G20G
Total Memory Required Footnote 2~ 2 GB~ 3 GB~ 6 GB~ 13 GB~ 34 GB

Footnote 1

  • The parameter sga_target should be used for Oracle 10g or 11g based environments such as Release 12. This replaces the parameter db_cache_size, which was used in Oracle 9i based environments. Also, it is not necessary to set the parameter undo_retention for 10g or 11g-based systems, since undo retention is set automatically as part of automatic undo tuning.
  • Enabling the 11g Automatic Memory Management (AMM) feature is supported in EBS, and has been found to be useful in scenarios where memory is limited, as it will dynamically adjust the SGA and PGA pool sizes. AMM is enabled by using the memory_target and memory_max_target initialization parameters. MEMORY_TARGET specifies the system-wide sharable memory for Oracle to use when dynamically controlling the SGA and PGA as workloads change. The memory_max_target parameter specifies the maximum size that memory_target may take. AMM has proven useful for small to mid-range systems as it simplifies both the configuration and management. However, many customers with large production systems have experienced better performance with manually sized pools (or large minimum values for the pools). On Linux, Hugepages has resulted in improved performance; however, this configuration is not compatible with AMM. For large mission-critical applications systems, it is advisable to set sga_target with a minimum fixed value for shared_pool_size and pga_aggregate_target.

Footnote 2

  • The total memory required refers to the amount of memory required for the database instance and associated memory, including the SGA and the PGA. You should ensure that your system has sufficient available memory in order to support the values provided above. The values provided above should be adjusted based on available memory so as to prevent paging and swapping.

General Notes on Table

  • "Development or Test instance" refers to a small instance used only for development or testing, with no more than 10 users.
  • The range of user counts provided above refers to active Oracle E-Business Suite users, not total or named users. For example, if you plan to support a maximum of 500 active Oracle E-Business Suite users, then you should use the sizing as per the range 101-500 users.
  • The parameter values provided in this document reflect a small instance configuration, and you should adjust the relevant parameters based on the Oracle E-Business Suite user counts as listed in the table above.
  • The "csp" and "no csp" options of the shared pool parameters refer to the use of cursor_space_for_time, which is documented in the common database initialization parameters section.

    Note: Enabling cursor_space_for_time can result in significantly larger shared pool requirements.

Comentarios

Publicar un comentario

Entradas populares de este blog

[ORACLE] Script Espacio ocupado y libre de tablespaces (incluyendo los temporary)

En este script utilizo la vista dba_temp_free_space que está disponible a partir de la versión 11g de Oracle: SELECT a . tablespace_name , a . asignado , b . libre , c . contents , c . status FROM ( SELECT tablespace_name , SUM ( bytes )/ 1024 / 1024 AS asignado FROM dba_Data_files GROUP BY tablespace_name ) a , ( SELECT tablespace_name , SUM ( bytes )/ 1024 / 1024 AS libre FROM dba_free_space GROUP BY tablespace_name ) b , ( SELECT tablespace_name , contents , status FROM dba_tablespaces ) c WHERE a . tablespace_name = b . tablespace_name (+) AND a . tablespace_name = c . tablespace_name UNION ALL SELECT d . TABLESPACE_NAME , d . ALLOCATED_SPACE / 1024 / 1024 , d . FREE_SPACE / 1024 / 1024 , e . contents , e . status FROM dba_temp_free_space d , dba_tablespaces e WHERE d . tablespace_name = e . tablespace_name AND contents = 'TEMPORARY' ; Para la versión
Leer más

Oracle - Apuntes sobre LOBs para DBAs

Imagen
Objetivo Este documento recopila información que considero interesante conocer sobre los LOBs (en la documentación oficial los llaman formalmente como “ Large Objects ”, y a partir de la 11.1 como “ Basic Files ”), sobre todo a la hora de la administración. Lógicamente no sustituye a la documentación oficial, sólo son unos apuntes que he ido recopilando de varias fuentes y he preferido juntarlo en un documento único. No hablo para nada de tablas particionadas ni de IOT´s. Esto en la documentación oficial. Introducción ¿Qué son? ¿Para qué se utilizan? ¿Clases de LOBs? ¿Características? Todo esto mejor en la documentación oficial, pero haré un resumen esquemático. Estructura Cuando se crea una tabla conteniendo una columna de tipo LOB, se crean dos segmentos para contener la columna LOB especificada. Estos segmentos son de tipo LOBSEGMENT y LOBINDEX . El segmento LOBINDEX se utiliza para acceder a las páginas/trozos (chunks) que se almacenan en el segmento LOBSEGMENT. Cuando cr
Leer más

Oracle - Traza 10046 (SQL_TRACE)

Cómo obtener una traza 10046 (SQL_TRACE) Hay varias formas de obtener una traza 10046 para el diagnostico de rendimiento en consultas. 1º Evento 10046 El código de evento especial es equivalente a establecer el parámetro de inicialización SQL_TRACE=true La ventaja de utilizar este evento es que posibilita detalles extra en el fichero de traza generado, dependiendo del nivel especificado en el evento. 10046 EVENT levels:         1  - Enable standard SQL_TRACE functionality (Default)         4  - As Level 1 PLUS trace bind values         8  - As Level 1 PLUS trace waits              This is especially useful for spotting latch wait etc.              but can also be used to spot full table scans and index scans.         12 - As Level 1 PLUS both trace bind values and waits         16 - Added in 11g to generate STAT line dumps for each execution Sintaxis de ejemplo:      A nivel de sesión:             alter session set events '10046 trace
Leer más