11. Database system files
The following database systems are enabled during Oracle database operations and are located on the database server. Note that data pieces are physical pieces that belong to the database container and will not be described here.
• control pieces: control pieces are required pieces, which store metadata about data pieces and online redo pieces; for example, their name and status. The database instance needs this information to open the database. The control file also contains metadata that must be accessible when the database is not open. It is strongly recommended that you make multiple copies of the control pieces in the database server to achieve the scalability.
Parameter ⽂ piece: this required piece defines the configuration ⽅ type when the database instance is started. It can be an initialization parameter piece (P piece) or a server parameter piece (SP piece).
Online redo records: these necessary records store the changes in the database when making changes and restore the data.
Dynamic diagnosis repository (ADR): ADR is a repository based on ⽂ pieces ⽤ to store database diagnosis data, such as tracking, dump, alarm ⽇ log, operation ⾏ status monitor report, etc. It has a unified record structure across multiple instances and products. Database, Oracle ASM, listener, Oracle clusterware and other Oracle products or components store all diagnostic data in ADR. Each instance of each product stores the diagnostic data in its primary record in ADR.
• backup ⽂ pieces: these optional ⽂ pieces ⽤ are recovered in the database. In general, you can restore a backup piece when the original piece is damaged or deleted due to a media failure or user error.
• archived redo records: these optional records contain the continuous history of data changes made by the database instance. With these backups and database backups, you can recover lost data. In other words, archive records can recover restored data.
Password piece: this optional piece enables users of SYSDBA, sysoper, sysbackup, sysdg, syskm, sysrac and sysasm to remotely connect to the database instance and perform management tasks.
• Wallet: for the large-scale deployment of the application to connect password credentials to the database, such credentials can be stored in the client Oracle wallet. Oracle wallet is a secure software container that stores authentication and signing credentials. Possible wallets include Oracle wallets based on user credentials, encryption wallets based on transparent data encryption (TDE) and Oracle public cloud (OPC) wallets based on database backup cloud module. The wallet is optional, but it is recommended to use the.
• block change tracking ⽂ pieces: block change tracking improves the performance of incremental backup by recording the changed blocks in the block change tracking ⽂ pieces. During incremental backup, Oracle Recovery Manager (RMAN) scans all data blocks to determine which blocks have changed, and enables this file to identify the changed blocks to be backed up. Block change tracking pieces are optional.
• flashback: Flashback Database is similar to regular point in time recovery in effect. It allows you to restore the database to a near state one time. Flashback Database makes its records record mechanism, creates flashback records and stores them in the fast recovery area. Flashback to the database is only possible when flashback records are OK. To use this feature, you must set up the database in advance to create flashback logs. Flashback is optional.
Control, online redo, and archive redo records can be multiplexed, which means that two or more identical copies can be dynamically maintained in a separate location.
Note: for more details, see the following: physical storage structures
12. Application containers
The application container is an optional CDB component created by the user, which stores the data and metadata of the application PDB. CDB can contain zero or more application containers. The application program container consists of application program root record and application program PDB, which are inserted into CDB root record. The application root record belongs to CDB root record, does not belong to other containers, and stores public metadata and data.
The typical application program will install the common users, metadata linked public objects and data linked public objects of the application program. You can create multiple sales related PDBs in the application containers, which share the application backend composed of group public table and table definition.
There are system, sysaux, temp, users and optional undo tablespaces in the application root, application type and application PDB. Each table space represents one or more data pieces.
Note: for more details, see the following: About Application containers
13. Automatic diagnostic repository (ADR)
Dynamic diagnostic repository (ADR) is a central repository for system wide tracking and recording of database diagnostic data. It includes the following:
Background trace pieces: each database background process can write associated trace pieces. When a process detects an internal error, it dumps information about the error into its trace file. Write some information of tracking files for database administrator to use, and other information for Oracle to use. Generally, the database background process trace file name contains the Oracle system identifier (SID), background process name and operating system process number. An example of a trace piece in a reco process is mytest reco 10355.trc.
• foreground trace pieces: each server process can write associated trace pieces. When a process detects an internal error, it dumps information about the error into its trace file. Server process tracking file name contains Oracle SID, string ora and operating system process number. An example of the server process trace file name is mytest \ ora \ 10304.trc.
• dump ⽂ pieces: diagnostic dump ⽂ pieces are ⼀ special tracking ⽂ pieces, which contain detailed time point information about status or structure. Dump ⽂ pieces are usually the ⼀ secondary output of diagnostic data in response to events, and ⽽ trace ⽂ pieces are often the continuous output of diagnostic data.
Operation status monitor report: the Oracle database includes a framework called operation status monitor, which runs diagnostic checks on the database. Operation condition check can detect damage of parts, physical and logical block, undo and redo, data dictionary, etc. The health examination will report its findings and, in many cases, provide recommendations to solve the problem.
Event package: for uploading the diagnostic data to the definition method held by Oracle, you first collect the data into an intermediate logical structure called event package. A package is a collection of metadata stored in the ADR and points to diagnostic data pieces and other pieces inside and outside the ADR. When creating a package, select one or more questions to add to the package. Then, the ⽀ console adds problem information, event information and diagnostic data (such as tracking ⽂ pieces and dumps) related to the selected problem to the package.
Event dump: when an event is sent, the database writes one or more dumps to the event record created for the event. The event dump also contains the event number in the piece name.
Alarm records: the alarm records of the database are the records of messages and errors in chronological order. Oracle recommends that you check the alarm logs regularly.
Note: for more details, see the following: automatic diagnostic repository
14. Backup files
Database backup can be physical backup or logical backup.
• a physical backup is a copy of a physical database. You can make a physical backup of recovery manager (RMAN) or operating system hypervisor.
• logical backups contain tables, stored procedures, and other logical data. You can enable Oracle database implementation programs (such as data pump exports) to extract logical data and store it in binary files. Logical backups can supplement physical backups.
Database backups created by RMAN are stored as image copies or backup sets.
• image copy is a bit by bit disk copy of data, control or archive redo records. You can have the operating system hypervisor or RMAN create an image copy of the physical artifacts and then have any artifacts restore them. Image copies are very useful to disks because you can incrementally update them and restore them in place.
• backup set is a proprietary format created by RMAN, which contains ⼀ or multiple data ⽂ pieces, archived redo ⽇ records ⽂ pieces, and data in control ⽂ pieces or server parameters ⽂. The ⼩ unit of the backup set is ⼀, which is called backup ⽂, in decimal ⽂. Backup set is the only form that RMAN can write backup to sequential devices (such as tape drives). The advantage of backup set is that RMAN can save the space of backup data pieces by compressing the blocks that are not compressed. The backup set includes only those blocks in the data file that store the data. The backup set can also be compressed, encrypted, sent to tape, and uncompressed at the same time.
RMAN can interact with media management library (MML), system backup to tape (SBT) software (backup can be created to tape), Oracle database backup cloud service, or zero data loss recovery device (commonly referred to as recovery device).
Note: for more details, see the following:
Backup and Recovery
About Zero Data Loss Recovery Appliance
15. Process monitoring process (PMON)
Process monitoring process (PMON) is a background process. It scans all processes regularly to find any abnormal dead processes. PMON then coordinates the cleanup performed by the cleanup master process (clmn) and the cleanup slave process (CLNN).
PMON is not running as a thread as an operating system process. In addition to the database instance, PMON can also run on the Oracle active storage management (ASM) instance and the Oracle ASM agent instance.
Note: for a complete list of processes, see the following: background processes
The article is being updated. Please look forward to the following~