Latch statistics
Job details
Name: | Latch statistics |
Platform: | Oracle |
Category: | Performance |
Description: | Collects latch status statistics. |
Long description: | Task collects latch status statistics. |
Version: | 1.6 |
Default schedule: | 5,15,25,35,45,55 * * * |
Requires engine install: | Yes |
Compatibility tag: | .[type=‘instance’ & databasetype=‘oracle’]/.[hasengine=‘YES’] |
Parameters
Name | Default value | Description |
---|---|---|
keep data for | 14 | The amount of days the statistics are kept in the history table. |
Job Summary for dbWatch Control Center Monitoring Job: Latch Statistics
The monitoring job configured within dbWatch Control Center is designed to collect statistics about Oracle database latches, focusing on performance aspects such as latch access efficiency metrics.
- “Purpose”: The job is created to collect and evaluate the state of various latches in an Oracle database environment, providing insights into potential performance bottlenecks.
- “Why”: Monitoring latch statistics is crucial for managing database performan
ce. Latch contention generally indicates areas where the database is spending a considerable amount of time waiting for resources, thereby impacting overall performance.
- “Manual checking”: Although specific SQL commands are not provided in the description, the general approach to manually checking this information would be querying performance views related to latches in Oracle, such as V$LATCH and V$LATCHHOLDER.
Implementation Details
Database objects and procedures are defined and created to facilitate the capture and storage of latch statistics.
- “Tables and Views”: Several tables (dbw_latchstat, dbw_latch_name, dbw_latchstat_histr) and views (e.g., dbv_get_all_rank, dbv_misses_all_rank) are created to hold latch performance data incrementally polled from dynamic performance views.
- “Stored Procedures”: Procedures are put in place for capturing live data from the Oracle instance and populating the pre-defined tables. They handle periodic updates and historical data maintenance.
- “Scheduled Reports”: Reports and visualizations are scheduled to fetch and present data, enabling periodic review without manual query execution.
Dependencies and Database Environment
Oracle-specific base views and dependency tables are integral to the job’s operation.
- “Dependencies”: Objects such as dbv_memory_rank and dbv_redo_latches suggest an in-depth collection mechanism focused on various latch types such as memory and redo logs.
- “Cleanup”: Robust cleanup mechanisms ensure that failure during data gathering does not leave orphaned entries or incomplete data captures.
Monitoring and Reporting
Comprehensive reporting mechanisms are built into the job for regular reviews of latch statistics.
- “Reports”: Generated reports provide data on latch activities, categorizing them by different performance metrics such as latch gets, misses, sleeps, and overall hit ratio.
- “Visualization”: Charts and tables effectively illustrate trends over time, making it easier for database administrators to spot anomalies or patterns that could indicate performance issues.
Scheduling and Automation
Routine executions are scheduled to ensure data is current and reflective of the most recent database state.
- “Execution Schedule”: The job is scheduled to run at specific intervals, generating real-time data necessary for responsive database performance tuning.
- “Automation Level”: High automation minimizes the manual overhead involved in tracking and analyzing latch statistics.