Understanding Metrics in Oracle 19c: A Practical Guide

Metrics in Oracle 19c provide essential insights into the performance and health of your database. By monitoring these metrics, database administrators (DBAs) can detect issues early, optimize resource usage, and ensure smooth database operations. This entry explores key metrics views such as V$METRIC, V$SYSMETRIC, and V$METRIC_HISTORY, provides examples of setting alert thresholds, and discusses the most useful types of metrics and their applications.

Key Metrics Views in Oracle 19c

Oracle 19c offers several dynamic views that store and present performance metrics. Understanding these views is crucial for effective database monitoring.

V$METRIC

V$METRIC provides real-time performance data collected at short intervals, typically every minute. This view is ideal for tracking the immediate performance of your database, offering granular insights into various system activities.

SELECT METRIC_NAME, VALUE, BEGIN_TIME, END_TIME 
FROM V$METRIC 
WHERE METRIC_NAME = 'CPU Usage Per Sec' 
ORDER BY BEGIN_TIME DESC;

This query retrieves recent CPU usage data, which can be crucial when diagnosing sudden spikes in resource usage.

V$SYSMETRIC

V$SYSMETRIC is similar to V$METRIC but typically aggregates data over slightly longer intervals, providing a broader view of system performance. It’s particularly useful for analyzing trends over time rather than minute-by-minute changes.

SELECT METRIC_NAME, VALUE 
FROM V$SYSMETRIC 
WHERE METRIC_NAME = 'Database Wait Time Ratio';

This query returns the current value for the Database Wait Time Ratio, helping identify potential I/O bottlenecks.

V$METRIC_HISTORY

V$METRIC_HISTORY stores historical metric data, allowing DBAs to analyze trends and compare performance over time. This view is essential for understanding how performance metrics evolve and identifying persistent issues.

SELECT METRIC_NAME, VALUE, BEGIN_TIME, END_TIME 
FROM V$METRIC_HISTORY 
WHERE METRIC_NAME = 'DB Time Per Sec' 
AND BEGIN_TIME > SYSDATE - 1
ORDER BY BEGIN_TIME;

This query retrieves historical data for the DB Time Per Sec metric from the last day, which is useful for analyzing database performance over time.

Setting Alert Thresholds for Metrics

One of the practical applications of monitoring metrics is setting alert thresholds. This allows you to automate the detection of potential issues before they become critical.

Oracle provides the DBMS_SERVER_ALERT.SET_THRESHOLD procedure to set these thresholds. Here’s an example:

BEGIN
   DBMS_SERVER_ALERT.SET_THRESHOLD(
      metrics_id           => DBMS_SERVER_ALERT.TABLESPACE_PCT_FULL,
      warning_operator     => DBMS_SERVER_ALERT.OPERATOR_GT,
      warning_value        => '80',
      critical_operator    => DBMS_SERVER_ALERT.OPERATOR_GT,
      critical_value       => '95',
      observation_period   => 1,
      consecutive_occurrences => 1,
      instance_name        => NULL,
      object_type          => DBMS_SERVER_ALERT.OBJECT_TYPE_TABLESPACE,
      object_name          => 'USERS' -- Replace with your tablespace name
   );
END;
/

This script sets a warning threshold when the USERS tablespace exceeds 80% capacity, and a critical alert at 95%. This helps in proactive management of tablespace usage, ensuring that space issues are addressed before they impact database operations.

Most Useful Types of Metrics

Certain metrics are particularly useful for maintaining a healthy database environment. Below are some key types:

1. CPU Usage Metrics

Metrics such as CPU Usage Per Sec help in monitoring how much CPU resources the database is consuming. High CPU usage can indicate inefficient queries or inadequate hardware resources.

2. I/O Metrics

Metrics like Database Wait Time Ratio and Physical Reads Per Sec are vital for understanding how efficiently the database is handling I/O operations. Persistent high wait times or read rates can suggest underlying storage issues or poorly optimized queries.

3. Memory Usage Metrics

Metrics such as Buffer Cache Hit Ratio are crucial for ensuring that the database is efficiently using its memory. A low hit ratio could indicate that the database is frequently reading data from disk rather than memory, which can slow down performance.

4. Session Activity Metrics

Monitoring metrics like Session Count helps in understanding the load on your database. An unusually high number of sessions might suggest a need to increase the capacity or optimize application connections.

Use Cases for Metrics

1. Diagnosing Performance Issues

When users report slow database performance, metrics such as DB Time Per Sec or CPU Usage Per Sec provide immediate insights into potential bottlenecks, allowing DBAs to quickly identify and resolve the issue.

2. Capacity Planning

By analyzing metrics like Tablespace Usage and Sessions Highwater, DBAs can predict when additional resources will be needed and plan capacity upgrades accordingly.

3. Proactive Monitoring

Setting up alert thresholds based on metrics ensures that DBAs are notified of issues before they escalate. For example, receiving an alert when CPU usage exceeds a certain threshold allows you to investigate and mitigate the issue before it affects users.

Conclusion

Metrics in Oracle 19c are invaluable tools for monitoring and maintaining the performance and health of your database. By leveraging views like V$METRIC, V$SYSMETRIC, and V$METRIC_HISTORY, and setting appropriate alert thresholds, DBAs can ensure their databases run smoothly and efficiently. Understanding and using these metrics effectively can make the difference between a well-tuned system and one that constantly battles performance issues.

• V$METRIC in Database Reference
• DBMS_SERVER_ALERT in PL/SQL Packages and Types Reference