1. Oracle Autonomous Health Framework User's Guide
  2. CHM Inline Analysis

CHM Inline Analysis

Inline Analysis is a new feature that will automatically execute on all systems where Autonomous Health Framework (AHF) is installed.

Traditionally, the operating system (OS) metrics collected by Cluster Health Monitor (CHM) or System Health Monitor (SHM) are stored directly in the Grid Infrastructure (GI) Base repository. However, this repository is limited in size-250 MB on HAS 19 and 500 MB on MAIN.

On large FASAAS systems, which may run between 30,000 and 40,000 processes, these size limitations mean that CHM OS data can only be retained for a few hours. As a result, the necessary OS data is frequently missing in most Service Requests (SRs), making it challenging to diagnose and resolve issues effectively.

With the CHM Inline Analysis enhancement, instead of preserving large volumes of raw CHM OS data, the system analyzes these OS metrics and stores only the summary analyzed data. The processed analysis files are significantly smaller than the raw data, allowing for several months of data retention within a 100 MB repository quota.

How it works

  • Collection and Analysis:

  • OS metrics are collected by CHM/SHM as usual.

    • Every hour, Oracle Trace File Analyzer (TFA) triggers Inline Analysis which processes the hourly raw chmosdata / shmosdata gzip files.
  • Storage: Only the analyzed results are kept in the AHF home data repository, dramatically reducing space consumption.
  • Execution and Security: Inline Analysis runs as the root user and is configured with specific resource constraints to avoid impact on system performance.

Benefits

  • Extended Retention: Analyzed data can be retained for several months rather than just a few hours.
  • Reduced Storage Footprint: Only 100 MB of repository space is required, even for large systems.
  • Improved Diagnostics: OS data will be available in more SRs, enabling more effective and timely troubleshooting.

Table 4-28 Feature Comparison: Traditional vs. Inline Analysis

FeaturePrevious ApproachWith Inline Analysis
Data StoredRaw CHM OS dataAnalyzed (summarized) data
Repository Size (typical)250 MB (HAS 19), 500 MB (MAIN)100 MB
Data RetentionHoursMonths
Impact on DiagnosticsData often missing in service requests (SRs)Data available for analysis