Oracle Performance Evolution: Comprehensive Analysis of Versions 10.2.4, 10.1 FP2, and 9.5 FP4

​This comprehensive analysis compares the performance evolution across three major HWA (Workload Automation) releases, demonstrating significant improvements in the latest 10.2.4 version while highlighting the progression from 9.5 FP4 through 10.1 to the current release. 

Executive Summary 

• HWA 10.2.4 delivers the best overall performance across all metrics
• Significant improvements in scheduling delays and system stability
• Consistent resource utilization optimization across versions
• Enhanced throughput capabilities in the latest release
  
  

Test Environment Consistency
All tests were conducted under similar conditions to ensure fair comparison:

Common Configuration:

• Workload: 600K+ jobs across 25 workstations and 10,151 job streams
• Peak Load: 2,400+ jobs/min during 10-minute bursts
• Extended Load: 1,400 jobs/min sustained for 4 hours
• Infrastructure: VMware + RHEL + Oracle 19c

Version-Specific Details:

• 10.2.4: VMware 6, RHEL 9.4, WebSphere Liberty 25.0.0.3
  
• 10.1: VMware ESXi 7.0.3, RHEL 8.3, WLP 23.0.0.1
  
• 9.5 FP4: Referenced as baseline comparison
  
  

Performance Metrics Comparison
​
🚀 Peak Load Performance (10-Minute Burst)

⚡ Extended Load Performance (4-Hour Sustained) 

🔧 Critical Operations Performance

Performance Evolution Analysis

Major Improvements in 10.2.4

🎯 Extended Load Handling: The most significant improvement is in sustained workload performance, where 10.2.4 shows:

• 67% better schedule delay vs 10.1 (16s vs 48s)
• 60% better schedule delay vs 9.5 FP4 (16s vs 40s)
• 31% better mirroring delay vs 10.1 (18s vs 26s)
• 47% better mirroring delay vs 9.5 FP4 (18s vs 34s)

🏆 CPU Optimization: During peak loads, 10.2.4 achieves:

• 20% lower MDM CPU usage vs 10.1 (43% vs 54%)
• Comparable CPU usage to 9.5 FP4 baseline

Areas of Concern

📈 Peak Load Mirroring: 10.2.4 shows higher mirroring delays during peak loads compared to both previous versions, suggesting this area needs attention in future releases.

💾 Disk Utilization: Higher disk usage in 10.2.4 indicates increased I/O demands, though this may be acceptable given the performance improvements in other areas.

Note: 10.2.4 MAKEPLAN timing needs verification as it appears inconsistent with other improvements

---

Recommendations
For Production Deployment
✅ Strongly Recommended: HWA 10.2.4 is ideal for environments with:

• High sustained workloads (extended periods of heavy job processing)
  
• CPU-constrained systems
  
• Requirements for consistent long-term performance

Migration Considerations
⚠️ Monitor Closely: 

• Peak load mirroring performance in high-burst scenarios
  
• Disk I/O capacity planning
  
• MAKEPLAN operation timing (verify configuration)

Performance Optimization
🔧 Focus Areas for 10.2.4:

• Peak load mirroring delay optimization
  
• Disk utilization efficiency improvements
  
• Critical operation timing validation

---

Conclusion
HWA 10.2.4 represents a significant step forward in workload automation performance, particularly excelling in sustained high-load scenarios that are typical of enterprise production environments. While some peak-load metrics show regression, the dramatic improvements in extended load handling and CPU efficiency make it the superior choice for most enterprise deployments.

​The evolution from 9.5 FP4 → 10.1 → 10.2.4 demonstrates consistent focus on enterprise scalability, with version 10.2.4 achieving the best balance of performance and resource utilization for production workloads. 

---

​Analysis based on comprehensive testing with 600K+ job workloads across enterprise-grade infrastructure

---

​
​Alessandro Tomasi
Senior Technical Lead
Alessandro is a Software Engineer. He has an academic background in Computer Engineering.
In particular in technology customer support and automation testing

---

​Marco Gangemi
Software Engineer
Marco comes from an academic background in Physics. He joined HCL in 2018.
Since then, he has specialized in DevSecOps activities. He’s recently started on Performance tests.

---