| Oracle® Database Backup and Recovery Advanced User's Guide 10g Release 2 (10.2) Part Number B14191-01 |
|
|
View PDF |
| Oracle® Database Backup and Recovery Advanced User's Guide 10g Release 2 (10.2) Part Number B14191-01 |
|
|
View PDF |
This section describes instance recovery, and how Oracle's Fast-Start Fault Recovery improves availability in the event of a crash or instance failure. It also offers guidelines for tuning the time required to perform crash and instance recovery.
This section contains the following topics:
Instance and crash recovery are the automatic application of redo log records to Oracle data blocks after a crash or system failure. During normal operation, if an instance is shut down cleanly (as when using a SHUTDOWN IMMEDIATE statement), rather than terminated abnormally, then the in-memory changes that have not already been written to the datafiles on disk are written to disk as part of the checkpoint performed during shutdown.
However, if a single instance database crashes or if all instances of an Oracle Real Application Cluster configuration crash, then Oracle performs crash recovery at the next startup. If one or more instances of an Oracle Real Application Cluster configuration crash, then a surviving instance performs instance recovery automatically. Instance and crash recovery occur in two steps: cache recovery followed by transaction recovery.
The database can be opened as soon as cache recovery completes, so improving the performance of cache recovery is important for increasing availability.
During the cache recovery step, Oracle applies all committed and uncommitted changes in the redo log files to the affected data blocks. The work required for cache recovery processing is proportional to the rate of change to the database (update transactions each second) and the time between checkpoints.
Periodically, Oracle records a checkpoint. A checkpoint is the highest system change number (SCN) such that all data blocks less than or equal to that SCN are known to be written out to the data files. If a failure occurs, then only the redo records containing changes at SCNs higher than the checkpoint need to be applied during recovery. The duration of cache recovery processing is determined by two factors: the number of data blocks that have changes at SCNs higher than the SCN of the checkpoint, and the number of log blocks that need to be read to find those changes.
Frequent checkpointing writes dirty buffers to the datafiles more often than otherwise, and so reduces cache recovery time in the event of an instance failure. If checkpointing is frequent, then applying the redo records in the redo log between the current checkpoint position and the end of the log involves processing relatively few data blocks. This means that the cache recovery phase of recovery is fairly short.
However, in a high-update system, frequent checkpointing can reduce runtime performance, because checkpointing causes DBWn processes to perform writes.
To minimize the duration of cache recovery, you must force Oracle to checkpoint often, thus keeping the number of redo log records to be applied during recovery to a minimum. However, in a high-update system, frequent checkpointing increases the overhead for normal database operations.
If daily operational efficiency is more important than minimizing recovery time, then decrease the frequency of writes to data files due to checkpoints. This should improve operational efficiency, but also increase cache recovery time.
The Fast-Start Fault Recovery feature reduces the time required for cache recovery, and makes the recovery bounded and predictable by limiting the number of dirty buffers and the number of redo records generated between the most recent redo record and the last checkpoint.
The foundation of Fast-Start Fault Recovery is the Fast-Start checkpointing architecture. Instead of conventional event-driven (that is, log switching) checkpointing, which does bulk writes, fast-start checkpointing occurs incrementally. Each DBWn process periodically writes buffers to disk to advance the checkpoint position. The oldest modified blocks are written first to ensure that every write lets the checkpoint advance. Fast-Start checkpointing eliminates bulk writes and the resultant I/O spikes that occur with conventional checkpointing.
With the Fast-Start Fault Recovery feature, the FAST_START_MTTR_TARGET initialization parameter simplifies the configuration of recovery time from instance or system failure. FAST_START_MTTR_TARGET specifies a target for the expected mean time to recover (MTTR), that is, the time (in seconds) that it should take to start up the instance and perform cache recovery. Once FAST_START_MTTR_TARGET is set, the database manages incremental checkpoint writes in an attempt to meet that target. If you have chosen a practical value for FAST_START_MTTR_TARGET, you can expect your database to recover, on average, in approximately the number of seconds you have chosen.
|
Note: You must disable or remove theFAST_START_IO_TARGET, LOG_CHECKPOINT_INTERVAL, and LOG_CHECKPOINT_TIMEOUT initialization parameters when using FAST_START_MTTR_TARGET. Setting these parameters interferes with the mechanisms used to manage cache recovery time to meet FAST_START_MTTR_TARGET. |
The maximum value for FAST_START_MTTR_TARGET is 3600 seconds (one hour). If you set the value to more than 3600, then Oracle rounds it to 3600.
The following example shows how to set the value of FAST_START_MTTR_TARGET:
SQL> ALTER SYSTEM SET FAST_START_MTTR_TARGET=30;
In principle, the minimum value for FAST_START_MTTR_TARGET is one second. However, the fact that you can set FAST_START_MTTR_TARGET this low does not mean that that target can be achieved. There are practical limits to the minimum achievable MTTR target, due to such factors as database startup time.
The MTTR target that your database can achieve given the current value of FAST_START_MTTR_TARGET is called the effective MTTR target. You can view your current effective MTTR by viewing the TARGET_MTTR column of the V$INSTANCE_RECOVERY view.
The practical range of MTTR target values for your database is defined to be the range between the lowest achieveable effective MTTR target for your database and the longest that startup and cache recovery will take in the worst-case scenario (that is, when the whole buffer cache is dirty). A procedure for determining the range of achievable MTTR target values, one step in the process of tuning your FAST_START_MTTR_TARGET value, is described in "Determine the Practical Range for FAST_START_MTTR_TARGET".
|
Note: It is usually not useful to set yourFAST_START_MTTR_TARGET to a value outside the practical range. If your FAST_START_MTTR_TARGET value is shorter than the lower limit of the practical range, the effect is as if you set it to the lower limit of the practical range. In such a case, the effective MTTR target will be the best MTTR target the system can achieve, but checkpointing will be at a maximum, which can affect normal database performance. If you set FAST_START_MTTR_TARGET to a time longer than the practical range, the MTTR target will be no better than the worst-case situation. |
To reduce the checkpoint frequency and optimize runtime performance, you can do the following:
Set the value of FAST_START_MTTR_TARGET to 3600. This enables Fast-Start checkpointing and the Fast-Start Fault Recovery feature, but minimizes its effect on runtime performance while avoiding the need for performance tuning of FAST_START_MTTR_TARGET.
Size your online redo log files according to the amount of redo your system generates. A good rule of thumb is to switch logs at most every twenty minutes. Having your log files too small can increase checkpoint activity and reduce performance. Also note that all redo log files should be the same size.
The V$INSTANCE_RECOVERY view displays the current recovery parameter settings. You can also use statistics from this view to determine which factor has the greatest influence on checkpointing.
The following table lists those columns most useful in monitoring predicted cache recovery performance:
Table 11-2 V$INSTANCE_RECOVERY Columns
| Column | Description |
|---|---|
TARGET_MTTR |
Effective mean time to recover (MTTR) target in seconds. This field is 0 if FAST_START_MTTR_TARGET is not specified. |
ESTIMATED_MTTR |
The current estimated mean time to recover (MTTR) in seconds, based on the current number of dirty buffers and log blocks. This field is always calculated, whether or not FAST_START_MTTR_TARGET is specified. |
For more details on the columns in V$INSTANCE_RECOVERY, see Oracle Database Reference.
As part of the ongoing monitoring of your database, you can periodically compare V$INSTANCE_RECOVERY.TARGET_MTTR to your FAST_START_MTTR_TARGET. The two values should generally be the same if the FAST_START_MTTR_TARGET value is in the practical range. If TARGET_MTTR is consistently longer than FAST_START_MTTR_TARGET, then set FAST_START_MTTR_TARGET to a value no less than TARGET_MTTR. If TARGET_MTTR is consistently shorter, then set FAST_START_MTTR_TARGET to a value no greater than TARGET_MTTR.
To determine the appropriate value for FAST_START_MTTR_TARGET for your database, use the following four step process:
The FAST_START_MTTR_TARGET initialization parameter causes the database to calculate internal system trigger values, in order to limit the length of the redo log and the number of dirty data buffers in the data cache. This calculation uses estimated time to read a redo block, estimates of the time to read and write a data block, as well as characteristics of typical workload of the system, such as how many dirty buffers corresponds to how many change vectors, and so on.
Initially, internal defaults are used in the calculation. These defaults are replaced over time by data gathered on I/O performance during system operation and actual cache recoveries.
You will have to perform several instance recoveries in order to calibrate your FAST_START_MTTR_TARGET value properly. Before starting calibration, you must decide whether FAST_START_MTTR_TARGET is being calibrated for a database crash or a hardware crash. This is a consideration if your database files are stored in a file system or if your I/O subsystem has a memory cache, because there is a considerable difference in the read and write time to disk depending on whether or not the files are cached. The appropriate value for FAST_START_MTTR_TARGET will depend upon which type of crash is more important to recover from quickly.
To effecitvely calibrate FAST_START_MTTR_TARGET, make sure that you run the typcial workload of the system for long enough, and perform several instance recoveries to ensure that the time to read a redo block and the time to read or write a data block during recovery are recorded accurately.
After calibration, you can perform tests to determine the practical range for FAST_START_MTTR_TARGET for your database.
To determine the lower bound of the practical range, set FAST_START_MTTR_TARGET to 1, and start up your database. Then check the value of V$INSTANCE_RECOVERY.TARGET_MTTR, and use this value as a good lower bound for FAST_START_MTTR_TARGET. Database startup time, rather than cache recovery time, is usually the dominant factor in determining this limit.
For example, set the FAST_START_MTTR_TARGET to 1:
SQL> ALTER SYSTEM SET FAST_START_MTTR_TARGET=1;
Then, execute the following query immediately after opening the database:
SQL> SELECT TARGET_MTTR, ESTIMATED_MTTR
FROM V$INSTANCE_RECOVERY;
Oracle responds with the following:
TARGET_MTTR ESTIMATED_MTTR 18 15
The TARGET_MTTR value of 18 seconds is the minimum MTTR target that the system can achieve, that is, the lowest practical value for FAST_START_MTTR_TARGET. This minimum is calculated based on the average database startup time.
The ESTIMATED_MTTR field contains the estimated mean time to recovery based on the current state of the running database. Because the database has just opened, the system contains few dirty buffers, so not much cache recovery would be required if the instance failed at this moment. That is why ESTIMATED_MTTR can, for the moment, be lower than the minimum possible TARGET_MTTR.
ESTIMATED_MTTR can be affected in the short term by recent database activity. Assume that you query V$INSTANCE_RECOVERY immediately after a period of heavy update activity in the database. Oracle responds with the following:
TARGET_MTTR ESTIMATED_MTTR 18 30
Now the effective MTTR target is still 18 seconds, and the estimated MTTR (if a crash happened at that moment) is 30 seconds. This is an acceptable result. This means that some checkpoints writes might not have finished yet, so the buffer cache contains more dirty buffers than targeted.
Now wait for sixty seconds and reissue the query to V$INSTANCE_RECOVERY. Oracle responds with the following:
TARGET_MTTR ESTIMATED_MTTR 18 25
The estimated MTTR at this time has dropped to 25 seconds, because some of the dirty buffers have been written out during this period
To determine the upper bound of the practical range, set FAST_START_MTTR_TARGET to 3600, and operate your database under a typical workload for a while. Then check the value of V$INSTANCE_RECOVERY.TARGET_MTTR. This value is a good upper bound for FAST_START_MTTR_TARGET.
The procedure is substantially similar to that in "Determining Lower Bound for FAST_START_MTTR_TARGET: Scenario".
Once you have determined the practical bounds for the FAST_START_MTTR_TARGET parameter, select a preliminary value for the parameter. Choose a higher value within the practical range if your concern is with database performance, and a lower value within the practical range if your priority is shorter recovery times. The narrower the practical range, of course, the easier the choice becomes.
For example, if you discovered that the practical range was between 17 and 19 seconds, it would be quite simple to choose 19, because it makes relatively little difference in recovery time and at the same time minimizes the effect of checkpointing on system performance. However, if you found that the practical range was between 18 and 40 seconds, you might choose a compromise value of 30, and set the parameter accordingly:
SQL> ALTER SYSTEM SET FAST_START_MTTR_TARGET=30;
You might then go on to use the MTTR Advisor to determine an optimal value.
Once you have selected a preliminary value for FAST_START_MTTR_TARGET, you can use MTTR Advisor to evaluate the effect of different FAST_START_MTTR_TARGET settings on system performance, compared to your chosen setting.
To enable MTTR Advisor, set the two initialization parameters STATISTICS_LEVEL and FAST_START_MTTR_TARGET.
STATISTICS_LEVEL governs whether all advisors are enabled and is not specific to MTTR Advisor. Make sure that it is set to TYPICAL or ALL. Then, when FAST_START_MTTR_TARGET is set to a non-zero value, the MTTR Advisor is enabled.
After enabling MTTR Advisor, run a typical database workload for a while. When MTTR Advisor is ON, the database simulates checkpoint queue behavior under the current value of FAST_START_MTTR_TARGET, and up to four other different MTTR settings within the range of valid FAST_START_MTTR_TARGET values. (The database will in this case determine the valid range for FAST_START_MTTR_TARGET itself before testing different values in the range.)
The dynamic performance view V$MTTR_TARGET_ADVICE lets you view statistics or advisories collected by MTTR Advisor.
The database populates V$MTTR_TARGET_ADVICE with advice about the effects of each of the FAST_START_MTTR_TARGET settings for your database. For each possible value of FAST_START_MTTR_TARGET, the row contains details about how many cache writes would be performed under the workload tested for that value of FAST_START_MTTR_TARGET.
Specifically, each row contains information about cache writes, total phyiscal writes (including direct writes), and total I/O (including reads) for that value of FAST_START_MTTR_TARGET, expressed both as a total number of operations and a ratio compared to the operations under your chosen FAST_START_MTTR_TARGET value. For instance, a ratio of 1.2 indicates 20% more cache writes.
Knowing the effect of different FAST_START_MTTR_TARGET settings on cache write activity and other I/O enables you to decide better which FAST_START_MTTR_TARGET value best fits your recovery and performance needs.
If MTTR Advisor is currently on, then V$MTTR_TARGET_ADVICE shows the Advisor information collected. If MTTR Advisor is currently OFF, the view shows information collected the last time MTTR Advisor was ON since database startup, if any. If the database has been restarted since the last time the MTTR Advisor was used, or if it has never been used, the view will not show any rows.
You can use the V$INSTANCE_RECOVERY view column OPTIMAL_LOGFILE_SIZE to determine the size of your online redo logs. This field shows the redo log file size in megabytes that is considered optimal based on the current setting of FAST_START_MTTR_TARGET. If this field consistently shows a value greater than the size of your smallest online log, then you should configure all your online logs to be at least this size.
Note, however, that the redo log file size affects the MTTR. In some cases, you may be able to refine your choice of the optimal FAST_START_MTTR_TARGET value by re-running the MTTR Advisor with your suggested optimal log file size.
