- Shopping Bag ( 0 items )
Prepare for the Oracle Certified Professional Oracle9i Database: New Features for Administrators exam—a required exam for upgrading your Oracle8i or earlier DBA certification—using this Oracle Press study guide. Throughout each chapter,you'll find in-depth details on all the material covered on this challenging exam,followed by practice questions and chapter summaries to highlight what you've learned. Authorized by Oracle Corporation,this book and CD-ROM package is the most comprehensive preparation tool available for the Oracle9i Database: New Features for Administrators exam.
Book and CD-ROM include more than 300 interactive exam-based questions. CD-ROM contains interactive exams that simulate the actual OCPexams and include hotlinked questions,answers,and explanations.
Some of the high-availability enhancements are
Oracle documents every change that is made to the database in the redo log. Changes are recorded in the redo log as redo entries, which are comprised of a set of change vectors. A change vector tracks a specific change that is made to a single data block. The redo logs also track the rollback segments in the database. Oracle uses the redo log to recover from a database crash. Oracle scans the redo logs, reads the change vectors, and applies them to the corresponding data block in the database.
To recover from a crash, you must recover the data blocks in the cache that were dirty at the time of the failure. The entries that pertain to the dirty blocks in the data cache are documented in the redo log. However, the redo log is very likely to contain entries for dirty data blocks that were written back to disk before the crash, and therefore were not dirty at the time of the failure. Oracle9i optimized the recovery process to exclude the redo log entries that do not apply to blocks that were dirty at the time of the failure (see Figure 2-1).
Oracle first does a quick pass through the redo logs to identify the change vectors that pertain to the dirty blocks and then stores this information in the PGA. In a second pass, Oracle uses the information in the PGA to apply just the identified changes. Since Oracle only sequentially reads the online log and does not actually access the data blocks themselves, the overall time to recover is minimized by the two-pass approach.
The time to recover is dependent on how recently the last checkpoint was done, since it determines the number of data blocks with a higher SCN than the checkpoint SCN. The time to recover is also dependent upon the number of redo log blocks that must be read to identify changes. Therefore, frequent checkpointing actions result in a lower time for instance recovery; the price for this benefit is the adverse impact on performance. This performance degradation occurs because checkpointing uses the database writer (DBWn) processes. You must make a trade-off between instance recovery time and operational database performance.
Oracle9i introduces the fast-start checkpointing architecture to implement fast-start instance recovery functionality. The fast-start checkpointing architecture performs incremental checkpointing, and it writes the oldest dirty blocks first to ensure that the most recent dirty blocks are associated with a higher SCN. You can now specify the target mean time to recover (MTTR) for the instance recovery by assigning the initialization parameter FAST_START_MTTR_TARGET a value between 0 and 3,600 seconds. When you set this parameter to 0, Oracle9i will not use this parameter to manage instance recovery time. The optimal value for FAST_START_MTTR_TARGET depends upon the database utilization, the system global area (SGA) size, and the Service Level Agreement (SLA) for that site. Your challenge is to optimize the FAST_START_MTTR_TARGET parameter such that it is in accordance with the SLA without an undue degradation in operational performance. You can verify checkpoint statistics with the Statspack and by querying the V$INSTANCE_RECOVERY view.
Oracle9i added the following three columns to V$INSTANCE_RECOVERY:
ALTER SYSTEM SET FAST_START_MTTR_TARGET = 240;
Oracle internally uses the FAST_START_MTTR_TARGET parameter value to calculate and set the FAST_START_IO_TARGET and LOG_CHECKPOINT_INTERVAL parameters. If you specify the FAST_START_IO_TARGET and LOG_CHECKPOINT_INTERVAL parameter values, they will override the values Oracle calculates from the FAST_START_MTTR_TARGET parameter.
Some parameters that you use to manage instance recovery are as follows:
Some applications where Oracle Flashback is useful are
Here is an example:
1. All the rows of the database are intact at 10:25; the database is at SCN#1.
2. The user accidentally deletes critical rows of information at 10:28; the database is at SCN#2.
3. The user recognizes this loss of information at 10:29.
4. At 10:30, the user flashes back to the database at SCN#1 before the information was deleted and retrieves the "missing" rows into a PL/SQL cursor.
5. At 10:32, the user returns to the current database by disabling the flashback and inserts the missing data from the PL/SQL cursor.
1. You must use automatic undo management as opposed to the older technique of rollback segments:UNDO_MANAGEMENT = AUTO
2. You must set the initialization parameter UNDO_RETENTION to specify how far back in seconds Oracle should retain undo information and therefore how far back you can perform a Flashback query. You can also update this parameter dynamically with the following statement:ALTER SYSTEM SET UNDO_RETENTION=1800
3. You must have EXECUTE privilege on the DBMS_FLASHBACK package....
|Pt. I||Oracle9i New Features Exam Guide|
|4||Performance and Scalability Enhancements||211|
|Pt. II||Oracle9i Certified Professional New Features Practice Exams|
|6||OCP Oracle9i Upgrade Exam: New Features for Administrators||289|
Posted June 20, 2003
This book has too many mistakes, even after checking the errata. Section on Advanced Replication is very poor. Many facts that are presented are incorrect. Do not expect much similarity with the Oracle 9i Upgrade Exam 1Z0-030.Was this review helpful? Yes NoThank you for your feedback. Report this reviewThank you, this review has been flagged.