Oracle Note--SGA Summary (ii)

Redo Log Buffers

Each time a user modifies the data, a related record is generated in the redo log buffer. Redo log buffer is a cyclic buffer that records the redo entries of user modification operations. LGWR caches the redo entries in the Redolog buffer buffer to the Redolog file in the database. The Redolog file is important for maintaining the integrity of the data in the database, and every time the database instance starts it checks whether the data file needs to be recovered and rolled back using the data recorded in the Redolog file. Because the distributed order of data in the Redolog file does not want the data distribution in the data file to be discrete, the LGWR data is synchronized much faster than DBWN. Once the data is synchronized to the Redolog file, it is considered that the operation of the data has been permanently saved.

The triggering synchronization conditions are as follows:

1. Once every 3 seconds

2. There is a submission request

3.LGWR Switch Log

4. Redo to full 1/3 or cached data exceeds 1M

The size of the Redo Log buffer is defined by the initialization parameter Log_buffer, which defaults to Max (512kb,128kb*cpu_count). Typically, this default value is sufficient.

　　
Shared pool

Shared pools cache various types of program data, such as execution plans, parsed SQL, system parameters, and data dictionaries. The structure looks like this:

　　

Library Cache:

This SQL statement is often cached by the user in the share SQL area after executing an SQL statement. When the user executes the SQL statement again, Oracle will go to the shared area to retrieve it. If an already parsed SQL statement exists, it will take advantage of the parsed results and avoid hard parsing to improve performance. The V$sql can query the cached Sql,v$sql_plan to cache the corresponding execution plan.

The SQL information cached in the library is invalidated when the following conditions occur:

1. Re-information statistics for tables, clusters, indexes, etc., which will regenerate the information in the SQL area to reflect the relevant statistics

2. Using DDL statements to modify table structure information

3. Changed the global database name.

4. Using the ALTER system flush Share_pool

Data dictitionary:

A data dictionary is a collection of table and view metadata that contains reference information about the structure of the database and its structure, users, and so on. The zone will be accessed frequently during Oracle parsing SQL statements.

Result Cache:

The result cache is distinguished between the SQL query result cache and the PL/s function cache.

The result of the SQL query results in a cached SQL query, and the query will return the query results directly from the result buffer when the next time it executes. Of course, if the cached data is inconsistent with the data in the table, the data in the table is modified, and the result data is automatically invalidated and then cached. The Result_cache_mode controls whether to cache all the query results or only the cache with Result_cache comments added.

The PL/SQL function results cache caches the results of each function execution, avoiding the same result function being called again for the same incoming parameter.
Flow Pool

Stores buffered queue information and provides memory for Oracle's stream capture process and application process.
Fixed SGA

An area of memory used by Oracle to manage internal transaction information. Includes database and instance status information, inter-process communication information. such as locks and so on.

Oracle Note--SGA Summary (ii)