5.2 Oracle Database 10g Self-Tuning and ManagementA major focus in Oracle Database 10g is the introduction of new self-tuning and management features in the database. Necessary for grid computing, these new features also make managing more familiar database environments much easier. Out of the box, the Oracle installation process builds an Intelligent Infrastructure. The infrastructure includes a workload repository (Automatic Workload Repository, or AWR) that is automatically populated with database statistical information gathered via "instrumentation" in the database code at very low overhead. Active session history is typically gathered every second, and the AWR is populated by default every 30 seconds. The Intelligent Infrastructure includes automatic setup of maintenance tasks that a DBA would normally schedule (such as ANALYZE to build statistics needed for the cost-based optimizer), server-generated alerts that occur just-in-time (not through a polling mechanism, as was present in previous EM deployment), and an advisory infrastructure.The new self-tuning and management features that DBAs and others can leverage include:Automatic Database Diagnostic Manager (ADDM) ADDM automatically tracks changes in database performance and makes recommendations on how to improve performance. Steps needed to resolve system utilization problems are typically reduced from watching events, exploring V$ tables, and identifying related SQL to simply reviewing the recommendations and accepting them. ADDM leverages data gathered in the workload repository. ADDM can be accessed through Advisor Central in EM. Automatic SQL Tuning Optimizer Rather than reviewing sessions with "bad SQL" by examining parameters such as system utilization and wait events, then observing Top Sessions in Enterprise Manager to identify such sessions and reviewing an explain plan, you can now simply review the ADDM recommendations, follow a link to SQL Tuning, and accept the recommendations. The SQL Tuning Optimizer performs a more extensive optimization investigation, and stores the results of this process in a Profile for automatic use by Oracle in the future. These Profiles can be used with packaged applications, as well as with your own custom code, and are used by the optimizer in subsequent resubmission of the SQL statements. Automatic Shared Memory Advisor The Memory Advisor is an expert system that eliminates manual adjustment of the buffer pool and shared pool sizes previously required to eliminate out-of-memory errors. Segment Advisor Use of the Segment Advisor eliminates the need to identify fragmented objects and then to use Enterprise Manager to reorganize the objects. The Segment Advisor advises which objects to shrink and allows you to simply accept the recommendations. Single Command Flashback Recovery The FLASHBACK DATABASE statement enables you to quickly revert the database to a state it was in at a past moment in time. Oracle Database 10g uses a log file containing data block images for changed blocks and other information. You can simply recover the table with a single FLASHBACK TABLE command. For more information on all of the new FLASHBACK commands, see Chapter 10. Automated Undo Management (AUM) Tuning AUM, the database feature that automatically allocates undo (rollback) space among active sessions, first appeared in Oracle9i. Oracle Database 10g AUM Tuning automatically determines optimal undo retention time based on system activity, and maximizes usage of the undo tablespace. Automated Storage Management (ASM) ASM provides a filesystem and volume manager in the database, enabling the automated striping of files and automating mirroring of database extents. DBAs simply define a pool of storage or disk group and manage the disk group through EM. Workloads can be dynamically rebalanced without taking the database down, so the DBA can add to the pool of storage while the database is available. ASM includes support of datafiles, log files, control files, archive logs, and RMAN and backup sets, and can be applied to a single system, or a RAC or grid computing environment. |