" Oracle large database systems combat Explanation on AIX / unix the " Discussion 32: Diagnosis and monitoring Oracle10g / 11g database performance methodology Wenping
Or continue the topic years ago, continue to discuss the diagnosis and monitoring methodology in the corresponding database performance Oracle10g / 11g. 10g / ADDM in 11g provides a good tool, but by no means universal, still in use. The correct methodology is particularly important, how to use it?
When an Oracle database performance problems occur applications must first be accurate and proper diagnosis, and making practical adjustments, to understand the position of the nodes and problems. Level of understanding of a system state history, current state, and with our own experience as a DBA database, which will help to quickly locate the problem, at least increase the likelihood of success to a large extent.
For an oracle system performance issues, analysis and concluded that the only way to get through statistical data. Oracle10g later version adds automatic Workload Repository (AWR) documented historical data on performance, corresponding to the function, Automatic Database Diagnostic Monitor (ADDM) is a AWR data analysis, based on Oracle's knowledge base and positioning performance Causes of labor issues, and propose recommended solutions. AWR role in this session is a historical performance data grab and save, ADDM AWR staff are analytical tools and data.
Specifically, each AWR snapshot, a trailing after ADDM analyzed, and the analysis results are stored in a database table. We can see the results by Oracle Enterprise Manager (OEM), or view the report directly through SQL * Plus.
In particular embodiments, performance tuning person ADDM first read output, and the analysis conclusion ADDM analysis. ADDM provides a relatively accurate database analysis, the so-called relatively accurate, refers to the technical content contained ADDM may in most cases more than you provide. ADDM has the following characteristics:
- Automatic Performance per hour automatically perform diagnostic reports
- It brings together Oracle performance diagnostic and tuning aspects of many years of wisdom
- Problem diagnosis quantify the time-based and provide accurate solutions
- Help identify the root cause, rather than a conclusion based on appearance
- Recommendations for the underlying problem, not the problem representation
- Database performance diagnostics minor impact on system performance, is negligible
ADDM powerful, but still based on a fundamental identity: an iterative process when performance tuning. The implication includes the following aspects: When you resolve a performance bottleneck, system performance problems may have emerged in another position, as in the following figure, we lock solution out of the first, second lock then revealed :
Seen from the results, improvements in the figure above, the direct result will be obvious, since there is a second lock play a role in the bottleneck. Therefore, for the use of ADDM, or any other optimization tools, one-time can not be expected to solve the problem, it will be an iterative process.
Automatic Database Diagnostic Monitor (ADDM) is a new version Oralcle user-friendly design, development of a typical case. After all, Oracle's engineers know yourself a little more. ADDM AWR provides any period of time during the sampling of statistical data analysis and provides the ability to adjust the program.
In Oracle10g, the confirming indicators on performance tuning, and different from previous versions of Oracle9i. Oracle10g proposed a comprehensive performance metrics: DB time. DB time is the cumulative time spent in the database server while processing a user request. It includes all active (non-idle) user session latency and CPU time. Performance analysis target is to reduce the DB time measurement time. DB time by V $ SESS_TIME_MODEL and V $ SYS_TIME_MODEL view query.
But the point to note, ADDM aims whole system is not the appropriate time to adjust individual sessions. If you prefer a single user session connection, end to end application you need to use tracking technology.
By reducing DB time, the database server load current reduction process working pressure, and further in the case of using the same resources, bear a greater load pressure. ADDM diagnosis report, a DB time of use of the arrangement according to the descending order, are arranged in front of the operation, the maximum likelihood system performance bottleneck.
ADDM report problems categorized as follows:
- Whether the system CPU usage by Oracle or other application constraints
- Operating system memory used as part of Oracle SGA, PGA size is adequate
- Performing I / O subsystem if there are problems
- SQL statement whether the system contains a high consumption
- Whether the system containing a high load of PL / SQL block, Java block
- What thermal block and the object (Real Application Cluster) present in the global cache structure RAC, and if the internal connection between its node communication problems
- Whether there are applications inefficient user connections, SQL parsing, lock problem
- Database configuration is problematic enough, whether improper log file size, checkpoints are too sparse or close etc.
- Under the existence of multi-user access concurrency issues, whether there is a cache block busy, race and other issues
- System hot objects, hot spots which SQL
ADDM labeling system to remove the problem, but also indicate what aspects of the system need to be considered in performance. Doing so saves the long time needed to spend on tuning engineers read the report, productivity increases.
In addition to diagnosing problems, ADDM also recommended possible solutions, these solutions may involve:
- Hardware-level adjustments: for example, increasing the CPU, memory, adjusting I / O subsystem
- Database level adjustments: for example a database system initialization parameters to adjust the placement
- Application Object Model adjustment: For example, the evolution of the partition table, management using ASSM tablespace
- Adjusting the application system: for example, the methods for changing the sequence number, and other technical bind variables
- Other consultants: for example, recommended the use of consultants brought Oracle10g tool to adjust the segment, SQL, restore and other operations
Unfinished, continued, see www.usedb.cn , Wenping ...
…
Reproduced in: https: //www.cnblogs.com/dbserver/articles/1680524.html