本文共 2670 字，大约阅读时间需要 8 分钟。

17608518: "row cache objects" Latch Contention in Oracle Data Guard

Problem Description

A critical issue (ID 17608518) has been identified in Oracle Data Guard, specifically in Active Standby (ADG) environments. This problem manifests as high row cache objects latch contention, characterized by wait events such as "latch: row cache objects" or "Child row cache objects". The issue primarily affects dc_rollback_segments, leading to performance degradation in read-intensive environments like Active Data Guard.

Affected Environments

• Product/Component: Oracle Server (RDBMS)
• Affected Versions: Versions BELOW 12.1 are believed to be affected.
• Confirmed Affected Versions: No specific versions confirmed yet.
• Platforms Affected: Generic, affecting all or most platforms.

Fix and Implementation

The provided fix introduces a latchless enhancement to alleviate row cache objects latch contention on dc_rollback_segments. This enhancement is specifically designed for read-only environments such as Active Data Guard, where row cache lookups are primarily for read operations. However, it is crucial to use this enhancement with caution in read-write environments, as latchless lookups may fail and revert to the latching mechanism, increasing overhead.

Enabling the Fix

event='17608518 trace name context forever, level 3'

ALTER SYSTEM SET event = '17608518 trace name context forever, level 3' SCOPE=spfileSID='*';

Technical Insights

The issue is closely related to Oracle Note ID 1550722.1, which provides detailed insights into the problem and its resolution. The problem is often diagnosed through monitoring tools that track latch wait events and contention metrics on dc_rollback_segments.

Key Observations

• High Contention: The issue is characterized by high row cache objects latch contention, indicating that the database is frequently accessing row cache objects, leading to contention.
• Impact on Operations: The problem can significantly impact the performance of data guard environments, particularly during peak read operations.
• Workaround: In cases where the fix cannot be applied immediately, consider monitoring the dc_rollback_segments and adjusting their configuration to reduce contention.

Conclusion

This fix addresses a critical performance issue in Oracle Data Guard environments. Proper implementation and configuration are essential to ensure optimal performance post-deployment. For further details and support, consult Oracle Support or refer to the provided reference note.