Data Guard Log Management Tips Oracle Database Tips by Donald BurlesonDecember 9, 2015

Oracle Data Guard - Log Management Services

The log management service is an integral part of Data Guard architecture. It is comprised of two components: Log Transport service and Log Apply service. The concept of Log Transport and Log Apply services are similar for a physical and logical standby database; however, there are differences at the detailed level of the working of log management for these two types of standby databases.

The processes involved and the architectural details of log transport and log apply services were provided in Chapter 2, "Data Guard Architecture". This chapter will mainly focus on the characteristics of these two services, influence of log transport service on the disaster recovery policies, and the implementation of various protection modes in Data Guard configuration.

In addition, it will briefly introduce the protection modes available in Release 1 of Oracle9i database for the benefit of a DBA running databases on Oracle9i Release 1.

Log Management Services - Log Transport Services

The log transport service is responsible for the successful transmission of redo data from the primary database to a Oracle instance over Oracle Net. In brief, the log transport service involves the log writer (LGWR) or archiver (ARCH or ARCn) process on the primary database site and the Remote File Server (RFS) on the Oracle instance for transmission of data changes on the primary database in the form of redo records.

The choice of log writer or archiver on the primary database as the log transfer agent depends on the data protection mode. The Oracle instance should be started and mounted in order for the log transport service to deliver archived redo log files to it. The DBA can opt to send either a complete archived redo log file or individual redo entries as these are generated on the primary database. For the latter, standby redo logs will have to be created on the standby site. The details on the method and form of redo transmission will be described in detail later in this chapter.

The log destination can be a physical standby database, a logical Oracle instance or an archive log repository without any database files. The archive log destination without an underlying database is used as an alternate repository for archived logs from the primary database.

In some configurations, a common archive log repository can be set up to cater to more than one Oracle instance as shown in Figure 5.1. In this scenario, the common archival destination receives archived redo logs from the primary database and transmits them to standby sites using Oracle Net. This setup can be very useful if the standby databases are remotely located over WAN, and the DBA does not want to put the performance of the primary database at risk while transferring archived redo logs.

Log Management Services - Log Transport Services

Figure 5.1 - Log Archival to a common archive destination

Data Guard Log Apply Services

The other part of log management is the log apply service. The log apply service reads archived log files from the directory specified by the standby_archive_dest initialization parameter and in the format specified by the log_archive_format parameter. The concept of log application is very different between the two types of standby databases.

On the physical standby database, the log apply service uses the block-per-block media recovery method to apply the changes. The logical Oracle instance utilizes the concept of LogMiner technology introduced in Oracle8i to convert the redo data into SQL statements and then applies those to the Oracle instance using SQL Apply operation. In the following section, the log apply services on physical and logical standby databases will be detailed.

Physical Standby Database

On the physical standby database, the log apply service reads and applies redo logs when the database is in managed recovery mode. In this mode, a process called the Managed Recovery Process (MRP), which takes care of the complete recovery process, is initiated by the database. The MRP can be started using the following statement:

ALTER DATABASE RECOVER MANAGED Oracle instance DISCONNECT FROM SESSION;

The RECOVER MANAGED Oracle instance statement was presented, in detail, in Chapter 3, Oracle instance Administration.

Start Logical Oracle instance apply tips