After an ASH report is generated, view its content to determine the possible causes of the instantaneous performance issues.
ASH report content
The ASH report records statistical data about the ASH report in the ASH Report section. This includes information such as the system version, operating system environment, information sampling time period, analysis time period, ASH and WR data sources, number of samples, number of events, and average number of active sessions. Based on this data, it generates information for 18 modules, including Top Active Tenants, Top Node Load, Top Groups, Top Foreground DB Time, Top Background DB Time, Top Sessions, Top IO Bandwidth, Top Blocking Sessions, Top Latches, Top DB Objects, Activity Over Time, Top Execution Phase, Top IO Events, Top SQL Statement Types, Top SQL with Top Events, Top SQL with Top Operator, Top PL/SQL Procedures, and Complete List of SQL Text.
These modules primarily display the performance data collected by the OBServer from two aspects: resource overhead and execution path overhead. Therefore, when analyzing specific performance issues, you can examine the relevant modules to determine whether the root cause is a resource bottleneck or inefficient execution.
ASH report
This section displays information about the sampling time, analysis time, and general statistics in the Active Session History (ASH) report.
- Cluster Name: The name of the cluster, which is the name of the database cluster.
- Observer Version: The version information of OceanBase AI Database, including the version number and detailed build version.
- Operation System Info: The operating system information, including the kernel version and architecture.
- User Input Begin Time: The time when sampling started, indicating the point in time when session activity recording began.
- User Input End Time: The time when sampling ended, indicating the point in time when session activity recording stopped.
- Analysis Begin Time: The time when analysis started, indicating the starting point in time for analyzing the sampled data.
- Analysis End Time: The time when analysis ended, indicating the ending point in time for analyzing the sampled data.
- Ash Data Source: The ASH data source. If ASH is not used, this column will be absent.
- Wr Data Source: The WR data source. If WR is not used, this column will be absent.
- Elapsed Time: The total number of seconds of all sampling times.
- Ash Num of Sample: The number of ASH samples, indicating the number of recorded session activities.
- Wr Num of Sample: The number of WR samples, indicating the number of performance snapshots generated within the analysis time range.
- Average Active Sessions: The average number of active sessions, indicating the average number of active sessions during the sampling period.
The statistics provide information about the activities, number of sessions, and system activity during the sampling period.
ASH Report
Cluster Name: test425
Observer Version: OceanBase 4.2.5.3 (203000012025022717-866087xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx)
Operation System Info: Linux(3.10.0-327.ali2019.alios7.x86_64)_x86_64
User Input Begin Time: 2024-09-22 10:26:47
User Input End Time: 2025-06-22 20:27:07
Ash Analysis Begin Time: 2025-03-05 12:59:50
Ash Analysis End Time: 2025-03-05 15:47:12
Wr Analysis Begin Time: 2025-02-28 09:43:43
Wr Analysis End Time: 2025-03-05 12:59:35
Ash Data Source: oceanbase.GV$ACTIVE_SESSION_HISTORY
Wr Data Source: oceanbase.DBA_WR_ACTIVE_SESSION_HISTORY
Elapsed Time: 453793
Ash Num of Sample: 11019
Wr Num of Sample: 475940
Average Active Sessions: 1.07
Top Active Tenants
This section breaks down resource consumption into foreground user requests and background tasks, allowing you to view the overall resource usage within a tenant.
- Tenant Name: The name of the tenant, identifying the information of a specific tenant.
- Session Type: The type of session. Possible values include BACKGROUND and FOREGROUND, representing background processes and foreground processes, respectively.
- Total Samples: The total number of records, indicating the total number of samples within the ASH report analysis time period.
- Wait Event Count: The number of wait events, indicating the number of samples where the session is waiting for an event.
- On CPU Samples: The number of samples on the CPU, indicating the number of samples where the session is executing on the CPU.
- Avg Active Sessions: The average number of active sessions, indicating the average number of database sessions that were active at any given time during the sampling period (either on the CPU or waiting for an event).
- % Activity: The percentage of activity, indicating the percentage of activity (CPU + wait) for a given tenant within the specified time period.
- Equivalent Client Load: The number of active connections between clients and the database during the ASH report analysis time period.
This information provides details about tenant activity, revealing different aspects of database performance and how tenants use resources.
In the Top Active Tenants module, performance issues may originate from within the database or in the interaction chain between the database and business applications. First, you need to identify the metrics Equivalent Client Load and Avg Active Sessions where the performance issue occurs. These two metrics are used to assess the impact of the database and links on performance. A larger ratio (Avg Active Sessions/Equivalent Client Load) indicates a greater impact of internal database overhead on performance; a smaller ratio indicates a greater impact of network links or the client.
For example, if the report shows a ratio of 88%, it indicates that the performance issue is mainly caused by database load. After confirming that there is a problem with the database load, you should further analyze the resource overhead within the database. In the Top Active Tenants module, foreground user requests consume 69.65% of database resources, while background tasks account for about 30%. Therefore, the performance bottleneck primarily occurs in the execution chain of user requests. Additionally, the wait events for foreground tasks consume a total of 1140 seconds of database time, while CPU execution consumes 72860 seconds. This indicates that the business model does not experience significant waiting due to resource limitations such as disk I/O or network bandwidth; the main overhead is concentrated on CPU execution. Since the business runs normally at the current version, it can be inferred that the cluster's CPU resources are sufficient. The slowdown in performance may be related to a decrease in execution efficiency at some stage of the SQL execution chain after the upgrade.

Top Node Load
This section displays the performance metrics of top nodes for the DB time metric.
IP: The IP address of the server.
Port: The port number of the server.
Session Type: The session type. Valid values include BACKGROUND and FOREGROUND, which indicate a background process and a foreground process, respectively.
Total Samples: The total number of records, indicating the total number of samples within the analysis time period of the ASH report.
Wait Event Samples: The number of wait events, indicating the number of samples where the session is waiting for an event.
On CPU Samples: The number of samples executed on the CPU, indicating the number of samples where the session is executing on the CPU.
Avg Active Sessions: The average number of active sessions, indicating the average number of database sessions that are active simultaneously during the sampling period at a given stage (either on the CPU or waiting for an event).
% Activity: The percentage of activity of events, indicating the percentage of activity (CPU + wait) of a given tenant within a specified time period.
Equivalent Client Load: The number of active connections between clients and the database. A larger ratio (Avg Active Sessions/Equivalent Client Load) indicates that internal database overhead has a greater impact on performance; a smaller ratio indicates that network links or the client have a greater impact on performance.
This information provides load-related performance metrics for each node in the database cluster.
This section displays the performance metrics of the top node for the DB time metric.

Top Groups
This section displays the resource consumer groups with high resource consumption.
Node: The node address.
Group Name: The name of the resource group.
Group Samples: The number of sampled session activity records in the current resource group.
% Activity: The percentage of activity of events, indicating the percentage of activity (CPU + wait) of a given tenant within a specified time period.
Program: The process name corresponding to the background session.
% Program: The proportion of the resource group (such as CPU or threads) occupied by this background session process.
Module: The name of the functional module in the program.
% Module: The proportion of this module among the active sessions of the program, used to locate hot operations within the program.
Action: The specific operation within the module.
% Action: The proportion of this operation among the active sessions of the module, further refining the granularity of performance analysis.
Avg Active Sessions: The average number of active sessions, indicating the average number of database sessions that are active simultaneously during the sampling period at a given stage (either on the CPU or waiting for an event).
Slot Begin Time: The start time of the recorded session activity period, in the format
YYYY-MM-DD HH:MM:SS(+ offset seconds), indicating the beginning moment of the statistical period.Slot Count: The total number of samples within the current time period. By default, ASH samples the active session state once per second. The Slot Count value equals the length of the time period, in seconds.
Action Key: Identifies the operation type that triggers active sessions. It is used to locate the resource usage of specific services or components and typically includes the following information:
- Source node: for example,
xx.xx.xx.xx:2882(IP address and port). - Tenant and resource group: for example,
tenant:1002/group:0. - Service type: for example,
T1002_LogService-LogRestoreService-RemoteLogWriter(Log Restore Service).
- Source node: for example,
Action Samples: The number of active sessions sampled with this operation (Action Key) within the current time period. A higher value indicates that this operation consumes more database resources.
This information provides top-level resource consumption groups and their key performance indicators, including the main performance events they encounter, the frequency of these events, the proportion of activity, and the average level of concurrent sessions.

Top Foreground DB Time
This section lists the main consumers of database foreground activity time by event type.
Node: The node address.
Event Name: The event that consumes the most DB Time, including both wait events and CPU-bound events.
Wait Class: The type to which the wait event belongs.
Event Samples: The number of session activity records sampled for each event.
Avg Active Sessions: The average number of active sessions, indicating how many database sessions were active at the same time during the sampling period (either on the CPU or waiting for events).
% Activity: The activity percentage of the event, representing the percentage of activity (CPU + wait) for a given tenant within the specified time period.
This information provides the top-ranked DB Time consumed by foreground sessions within the specified analysis period.
For example, from the Top Foreground DB Time data, it can be seen that lock conflict waits indeed occur during business task execution due to accessing transaction contexts. However, this overhead accounts for less than 1% overall, and no wait events corresponding to transaction commit (tx committing wait) were found among the wait events. Therefore, two performance-related hypotheses about the Top Active Tenants can be ruled out.
In summary, it can be confirmed that the performance degradation is not caused by resource bottlenecks. It is very likely that there has been a performance regression in the execution path of the new version.

Top Background DB Time
This section shows the DB Time generated by background sessions.
Node: The node address.
Program: The process name corresponding to the background session.
Module: The name of the functional module within the program.
Action: The specific operation within the module.
Event Name: The event that consumes the most DB Time, including both wait events and CPU-bound events.
Wait Class: The type of the event being waited for.
Event Samples: The number of session activity records sampled for each event.
% Activity: The activity percentage of the event, representing the percentage of activity (CPU + wait) for a given tenant within the specified time period.
Avg Active Sessions: The average number of active sessions, indicating how many database sessions were on average active during that phase during sampling (either on the CPU or waiting for an event).
This information provides insights into the most resource-intensive background processes, the main types of events they experience and their frequency of activity, their contribution to overall DB Time, and the average level of concurrent active sessions.

Top Sessions
This section displays information about the top sessions.
Session ID: The sampled session ID.
Program: The process name corresponding to the background session.
% Activity: The percentage of event activity, representing the percentage of activity (CPU + wait) for a given tenant within a specified time period.
Avg Active Sessions: The average number of active sessions, indicating how many database sessions were on average active during the sampling period at that stage (either on the CPU or waiting for events).
Event Name: The event that consumes the most DB Time, including both wait events and CPU-bound events.
Wait Class: The type to which the wait event belongs.
% Event: The proportion of total database activity accounted for by this event. A high percentage may indicate that this event is the main bottleneck in system performance.
SQL ID: The unique identifier of the SQL query.
Plan Hash: The value of the current SQL execution plan. Plans with the same hash value are identical.
% SQL ID: The proportion of total database activity accounted for by the SQL statement associated with this event. A high percentage indicates that this SQL is the primary source of resource consumption.
Sql Executions: The number of times the SQL statement was executed during the analysis period. Frequent execution may cause resource contention or lock conflicts.
This information provides statistics sorted by session for identifying active sessions. By comparing sampled session IDs, percentages of active sessions, events, event counts, event percentages, and the current user's active status, you can easily identify problematic sessions.

Top IO Bandwidth
This section lists the throughput of the top I/O operations from the perspectives of SQL and background tasks.
Node: The node address.
Program Module Action/SQL ID: The type of background task (such as log aggregation, transaction log writing) / the unique identifier of the SQL query, used to track the resource consumption of specific SQL statements.
Plan Hash: The value of the current SQL execution plan. Plans with the same hash value are identical.
Type: The type of I/O operation, distinguishing between read and write workloads.
IOPS: The number of I/O requests per second, reflecting the load pressure on storage devices. High IOPS may indicate frequent small-data-volume operations.
IO Size(MB): The total amount of data read or written by the task or SQL during the sampling period (MB). Background tasks typically involve larger data volumes (such as batch log writes).
IO Bandwidth(MB/s): The I/O throughput per second, measuring the actual utilization of storage bandwidth. Low bandwidth may be caused by data skew or hardware limitations.
Object ID: The primary data shard involved in the I/O operation. No data indicates no data skew or no samples collected during the sampling period.
% Object ID: The I/O percentage of the leader shard.
This information provides parameters such as Node, SQL ID, and IOPS to analyze the I/O throughput of SQL statements and background tasks. Combined with read/write types, data volume, and distribution characteristics, it helps identify performance bottlenecks.

Top Blocking Sessions
This section displays blocking session issues in the database caused by locks (enqueues), latches (latches), and buffer busy.
Blocking Session ID: The identifier of the session that blocks other sessions, used to locate the source of the problem.
% Activity: The percentage of activity of events caused by the blocking session. A high percentage indicates that this session is a major contention source.
Avg Active Sessions: The average number of active sessions of the blocking session during the sampling period, indicating how many database sessions were on average active at this stage during sampling (either on the CPU or waiting for events).
Holder User: The user holding the lock or resource.
Holder TX ID: The transaction ID holding the resource.
Holder SQL ID: The identifier of the SQL statement holding the resource, used to associate specific operations.
Event Caused: The type of wait event triggered by the blocking session.
% Event: The proportion of this event among similar blocks, helping identify high-frequency contention types.
XIDs: The associated transaction ID chain, used to track dependencies between transactions.
Top Waiting SQL ID: The identifier of the SQL statement most severely affected by the block.
% SQL ID: The proportion of this SQL in total blocking waits, used to prioritize optimization of high-impact queries.
This information provides fields such as Blocking Session ID and Holder TX ID to accurately locate the source of the block. Combined with the quantitative contention impact of % Activity and % Event, it uses Holder SQL ID and Top Waiting SQL ID to associate SQL-level root causes, and leverages XIDs to implement transaction chain tracing. Ultimately, it provides full-link diagnostic capabilities for database performance bottlenecks, from sessions and transactions to SQL.

Top Latches
This section displays the most heavily contested latches in the database.
Latch Wait Event: The name of the latch wait event.
Event Samples: The number of session activity records sampled for each event.
% Activity: The activity percentage of the event, representing the percentage of activity (CPU + wait) of a given tenant within a specified time period.
Avg Active Sessions: The average number of active sessions, indicating how many database sessions were on average active at this stage during sampling (either on the CPU or waiting for events).
This information provides the latches causing the highest waits, their wait event counts, activity percentages, and average number of active sessions, directly pointing to the key bottlenecks in database concurrency control. This information is crucial for diagnosing and resolving latch contention issues, reducing wait times, improving system response speed, and overall performance.

Top DB Objects
This section quantifies the access load and associated SQL of database objects (such as tables and partitions) by analyzing four key wait events: Application/Cluster/User I/O and buffer busy. This helps identify high-contention hotspots.
Node Address: The node address (format: IP:Port), used to identify the location of a physical node in a distributed environment.
Tenant ID: The unique identifier of a tenant.
Object ID: The tablet_id of a database table, used for metadata association.
Avg Active Sessions: The average number of active sessions, indicating how many database sessions were active at the same time during the sampling period (either on the CPU or waiting for events).
% Activity: The activity percentage of an event, representing the percentage of activity (CPU + wait) for a given tenant within a specified time period.
Execution Count: The total number of SQL executions for this object. Frequent access may trigger resource contention.
Event: The corresponding top wait event (up to 5 events are printed).
% Event: The percentage of the corresponding top wait event (up to 5 events are printed).
SQL ID: The SQL ID with high load in the wait event.
% SQL ID/Module: The percentage of SQL IDs or modules with high load in the wait event, used to prioritize optimization of high-impact queries.
Object Name (Type/Partition Name): The name, type, and partition information of the accessed database object.
This information quantifies the access load, associated SQL, and wait events of database objects (such as tables and partitions) to identify hotspot objects causing high resource contention, providing direct evidence for optimizing table structure, index design, and SQL performance.

Activity Over Time
This section displays the dynamic database activity within the analysis period using time interval information.
Slot Begin Time: The start time of the current time slot. Each time slot ends when the next one begins.
Slot Count: The total number of samples taken within the current time period. By default, ASH samples active session states once per second. The Slot Count value equals the length of the time period, in seconds.
Event Name: The event that consumes the most DB Time, including both wait events and CPU-bound events.
Wait Class: The type to which the wait event belongs.
Event Samples: The number of session activity records sampled for each event.
% Activity: The activity percentage of an event, representing the percentage of activity (CPU + wait) for a given tenant within a specified time period.
Avg Active Sessions: The average number of active sessions, indicating how many database sessions were active at the same time during the sampling period (either on the CPU or waiting for events).
This information provides a time-series overview of database activity dynamics during the analysis period, detailing the distribution of key events, activity intensity, and resource contention over time. This is invaluable for identifying performance issue trends, managing peak periods, and formulating time-based performance optimization strategies.

Top Execution Phase
This section displays the activities with a high proportion in the execution phase, categorized by session type, covering multiple aspects such as SQL execution, PL/SQL processing, and storage read/write.
Session Type: The session type. Possible values include BACKGROUND and FOREGROUND, representing background and foreground processes respectively.
Phase of Execution: The name of the execution phase, for example, IN_SQL_EXECUTION (SQL execution), IN_COMMITTING (commit process), IN_PLSQL_EXECUTION (PL/SQL execution), etc.
Active Samples: The number of occurrences or records in each execution phase.
% Activity: The percentage of activity for the event, representing the percentage of activity (CPU + wait) of a given tenant within a specified time period.
SQL ID: The SQL ID with high load among the waiting events.
% SQL ID/Module: The percentage of SQL IDs or modules with high load among the waiting events, used to prioritize optimization of high-impact queries.
Since performance issues with the base path typically affect all requests in the database, the Top Execution Phase provides a direct summary of the performance impact on the base path.
For example, from the content of the Top Execution Phase module, we can obtain the following information:
IN_SQL_EXECUTION accounts for 69.65% of the overhead. This part represents the total cost when SQL enters execution after obtaining the plan. Among the several execution phases defined in OceanBase AI Database, IN_SQL_EXECUTION includes IN_STORAGE_WRITE, IN_STORAGE_READ, and IN_RPC_ENCODE. This metric shows that the main cost of SQL lies in the execution (EXECUTION) phase, rather than the parsing (PARSER) or plan cache (PLAN CACHE) phase.
IN_STORAGE_WRITE is a sub-phase within the IN_SQL_EXECUTION phase, and this part of the cost accounts for 68.19%, indicating that the main performance cost occurs during data writes to the memtable.
IN_deadLOCK_ROW_REGISTER is a sub-phase of IN_STORAGE_WRITE, and this part of the cost accounts for 65.10%, indicating that most of the cost in this scenario occurs at this stage.
The above three metrics typically do not exceed 30% in import-type task benchmarks because, in import-type tasks, background tasks such as the entire system's memtable minor compaction, clog synchronization, and persistence to disk consume some CPU resources, while stages like SQL parsing and expression evaluation also consume CPU resources. In this task, however, the IN_STORAGE_WRITE phase accounts for nearly 70%, far exceeding the benchmark reference value, indicating that the performance issue in this business scenario primarily occurs at this stage. Additionally, the IN_DEADLOCK_ROW_REGISTER phase is mainly responsible for registering the row lock information held in the written data with the Deadlock Manager, facilitating background tasks to detect whether active transactions are deadlocked. Typically, this operation only updates and maintains a hashmap value, which is a very fast operation. In benchmarks, the cost for this phase usually does not exceed 3%. Clearly, in this system, this part of the cost is much higher than the empirical benchmark value.
Based on the analysis above, we can almost certainly determine that the performance issue lies in the IN_DEADLOCK_ROW_REGISTER phase. This is a general path for memtable writes. Performance issues that occur on this general path may affect either a specific SQL statement or the performance of all SQL statements accessing this path. So, in this problem, how do we determine the scope of affected SQL statements?
In addition to listing the proportionate costs of each phase predefined by OceanBase AI Database, the Top Execution Phase section also displays the SQL statement with the highest cost for each phase and its specific proportion. From the report data, we know that the SQL statement with the highest cost in the IN_DEADLOCK_ROW_REGISTER phase is D5B2150EA75EB405C25592A65708DD21. According to the text information displayed in the report, this SQL statement is an INSERT statement:
INSERT INTO TRADE_BASE_000_LOAD_42
SELECT TRADE_NO, TRADE_NO
FROM TRADE_BASE_000_LOAD PARTITION(p42)
WHERE ROWNUM <= 50000000;
The cost of this SQL statement in the IN_DEADLOCK_ROW_REGISTER phase is 0.67%, which is significantly different from the total cost of the entire phase, indicating that the total cost of the IN_DEADLOCK_ROW_REGISTER phase is not concentrated on a single SQL statement but is distributed across many SQL statements.

Top IO Events
This section provides statistics on I/O wait events for SQL statements and background tasks, identifying the sources of disk I/O bottlenecks, including high-load SQL, types of background tasks, and specific device performance metrics.
Node: The node address.
Program Module Action/SQL ID: The type of background task (such as log aggregation, transaction log writing) / the unique identifier of the SQL query, used to track the resource consumption of specific SQL statements.
Plan Hash: The numerical value of the current SQL execution plan. Identical plans will have the same hash value.
IO Event Samples: The total number of I/O wait events sampled during task execution.
%IO Event Samples: The proportion of the total number of I/O wait events among all samples, reflecting the frequency of I/O requests.
Top Event: The name of the primary I/O wait event.
IO Type: The type of the I/O wait event (read/write/asynchronous), used to distinguish operation modes.
% EVENT: The activity percentage of the I/O wait event.
Enqueue Time(S): The total time spent queuing for I/O requests, in seconds, reflecting device congestion.
Device Time(S): The total time spent executing I/O requests on the device, in seconds, measuring disk performance.
Callback Time(S): The total time spent on callback execution for I/O requests, in seconds.
This information quantifies I/O wait events for SQL and background tasks, combined with device enqueue time and execution time, to identify high-load SQL (by SQL ID and Plan Hash) and disk performance bottlenecks, providing a basis for optimizing I/O-intensive operations.

Top SQL Statement Types
This section shows the execution statistics summarized for various statement types during the ASH report analysis period, helping to analyze which statement type is causing business performance issues.
SQL Statement Type: Lists the most frequently executed SQL statement types, such as SELECT, UPDATE, INSERT, or DELETE.
Total Samples: The total number of executions of the given SQL statement type during the entire ASH report analysis period.
% Activity: The activity percentage of the event, representing the percentage of activity (CPU + wait) of the given tenant within the specified time period.
Sampled Executions: The number of sampled executions of this type of SQL (multiple executions may be independently counted due to different parameters).
Node: The node address.
% Node: The activity proportion of this type of SQL on the node, used to evaluate node load balancing.
% On CPU: The time proportion of this type of SQL running on the CPU. A high value indicates it is compute-intensive (such as complex aggregation).
% Event: The proportion of this event in the total database activity. A high percentage may indicate that this event is the main bottleneck for system performance.
This information provides categorized statistics on SQL statement execution, revealing the impact of different operation types (such as query, update) on database resource consumption and system activity. These data are valuable for identifying high-load SQL types, optimizing query strategies, balancing resource allocation, and performing targeted performance tuning.

Top SQL with Top Events
This section shows the SQL statements most closely associated with specific top events.
SQL ID: The SQL ID with high load among the wait events.
Plan Hash: The numerical value of the current SQL execution plan. Plans with the same hash value will have the same hash.
Active Samples: The number of occurrences or records per execution phase.
% Activity: The percentage of activity for the event, representing the percentage of activity (CPU + wait) of a given tenant within a specified time period.
Sampled Executions: The number of sampled executions for this type of SQL (multiple executions may be independently counted due to different parameters).
Top Event: The name of the primary I/O wait event.
% Event: The proportion of total database activity for this event. A high percentage may indicate that this event is the main bottleneck in system performance.
Top Operator/ExecPhase: The operation that takes the longest time in the current SQL execution plan.
% Operator/ExecPhase: The percentage of the total SQL execution time occupied by this operation/phase, used to locate execution plan bottlenecks.
SQL Text: The text of the SQL query.
This information identifies specific events that cause the highest proportion of activity during SQL statement execution, directly pointing to the core of the performance bottleneck. By identifying these "bottleneck" events and related SQL statements, database administrators can optimize SQL statements, adjust execution plans, or improve system configurations to reduce wait times and improve resource utilization.
The Top SQL with Top Events module displays information about specific SQL statements that consume more than 1% of the total time, but this section is empty. This indicates that the performance issue does not occur in any particular SQL statement; rather, overall SQL execution has become slower. Based on the report information, we have essentially identified that the performance issue is related to deadlock detection (IN_DEADLOCK_ROW_REGISTER) and is a general performance problem.
Top SQL with Top Events:
- This Section lists the SQL statements that accounted for the highest percentages event.
- Plan Hash: Numeric representation of the current SQL plan
- Active Samples: num of samples for top current SQL
- % Activity: activity percentage for given SQL ID
- Sampled Executions: represents the number of times the current SQL execution has been sampled
- Top Event: top event name for current SQL plan
- % Event: activity percentage for current SQL plan
- Top Operator/ExecPhase: top operator name or execution phase for current event
- % Operator/ExecPhase: activity percentage for given operator
+----------------------------------------+--------------------+--------------------+--------------+--------------------+----------------------------------------------------------------+--------------+--------------------------------------------------------------------------------------------------------------------------------+------------------------+----------------------------------------------------------------+
| SQL ID| Plan Hash| Active Samples| % Activity| Sampled Executions| Top Event| % Event| Top Operator/ExecPhase| % Operator/ExecPhase| SQL Text|
+----------------------------------------+--------------------+--------------------+--------------+--------------------+----------------------------------------------------------------+--------------+--------------------------------------------------------------------------------------------------------------------------------+------------------------+----------------------------------------------------------------+
+----------------------------------------+--------------------+--------------------+--------------+--------------------+----------------------------------------------------------------+--------------+--------------------------------------------------------------------------------------------------------------------------------+------------------------+----------------------------------------------------------------+
Top SQL with Top Operator
This section displays SQL statements that use specific operators in their execution plans and account for a high proportion of sampled session activity.
SQL ID: The SQL ID with high load in the wait events.
Plan Hash: The hash value of the current SQL execution plan, used to uniquely identify the SQL execution plan.
Active Samples: The number of occurrences or records per execution phase.
% Activity: The percentage of activity for the event, representing the percentage of activity (CPU + wait) of a given tenant within a specified time period.
Sampled Executions: The number of sampled executions for this type of SQL (multiple executions may be independently counted due to different parameters).
Top Operator: The operation that takes the longest time in the current SQL execution plan.
% Operator: The percentage of the total SQL execution time occupied by this phase, used to locate execution plan bottlenecks.
Top Event: The name of the primary I/O wait event.
% Event: The proportion of total database activity for this event. A high percentage may indicate that this event is the main bottleneck in system performance.
SQL Text: The text of the SQL query.
This information provides SQL operators associated with the highest proportion of activity, offering precise guidance for database administrators and developers to optimize SQL execution paths.
Top SQL with Top Operator:
- This Section lists the SQL statements that accounted for the highest percentages of sampled session activity with sql operator
- Plan Hash: Numeric representation of the current SQL plan
- Active Samples: num of samples for top current SQL
- % Activity: activity percentage for given SQL ID
- Sampled Executions: represents the number of times the current SQL execution has been sampled
- Top Operator: top operator name for current SQL plan
- % Operator: activity percentage for given operator
- Top Event: top event name for current operator
- % Event: activity percentage for given event
+----------------------------------------+--------------------+--------------+--------------+--------------------+--------------------------------------------------------------------------------------------------------------------------------+--------------+----------------------------------------------------------------+--------------+----------------------------------------------------------------+
| SQL ID| Plan Hash|Active Samples| % Activity| Sampled Executions| Top Operator| % Operator| Top Event| % Event| SQL Text|
+----------------------------------------+--------------------+--------------+--------------+--------------------+--------------------------------------------------------------------------------------------------------------------------------+--------------+----------------------------------------------------------------+--------------+----------------------------------------------------------------+
+----------------------------------------+--------------------+--------------+--------------+--------------------+--------------------------------------------------------------------------------------------------------------------------------+--------------+----------------------------------------------------------------+--------------+----------------------------------------------------------------+
Top PL/SQL Procedures
This section displays information about the top PL/SQL stored programs.
PLSQL Entry Subprogram: Displays the top-level PL/SQL entry subprogram of the application, which can be a procedure, function, trigger, or package initialization.
% Activity: The percentage of event activity, representing the percentage of activity (CPU + wait) for a given tenant within a specified time period.
PLSQL Current Subprogram: Displays the PL/SQL subprogram currently in execution, specifically the one being executed at the sampling point. If the value is
SQL, it indicates the percentage of time spent executing SQL statements within a particular PL/SQL entry subprogram; if the value is--, it indicates no specific subprogram is being executed.% Current: The proportion of time spent in the current subprogram (PL/SQL Current Subprogram) relative to the entry subprogram. If the current subprogram is SQL, it reflects the proportion of time spent executing the SQL.
This information provides details about the execution of the top PL/SQL stored programs, including information related to the top-level entry subprogram and the currently executing subprogram, along with their activity percentages.
Top PL/SQL Procedures:
- "PL/SQL Entry Subprogram" represents the application's top-level entry-point(procedure, function, trigger, package initialization) into PL/SQL.
- "PL/SQL Current Subprogram" is the pl/sql subprogram being executed at the point of sampling. If the value is "SQL", it represents the percentage of time spent executing SQL for the particular plsql entry subprogram.
- "PL/SQL Entry Subprogram" represents the application's top-level subprogram name
+------------------------------------------------------------+--------------------+------------------------------------------------------------+--------------------+
| PLSQL Entry Subprogram| % Activity| PLSQL Current Subprogram| % Current|
+------------------------------------------------------------+--------------------+------------------------------------------------------------+--------------------+
| oceanbase.dbms_workload_repository.ASH_REPORT| 0.00%| oceanbase.dbms_workload_repository.ASH_REPORT_TEXT| 0.00%|
+------------------------------------------------------------+--------------------+------------------------------------------------------------+--------------------+
Complete List of SQL Text
This section displays a list of SQL text, containing the SQL ID and text of the SQL queries.
SQL ID: The unique identifier of the SQL query.
SQL Text: The text of the SQL query.
This information provides a list of SQL queries, which can be used to identify, analyze, and optimize query performance.

Related wait events
Refer to Common wait events to understand the meaning of each wait event. This helps you quickly pinpoint issues while analyzing the report.