Challenge
A major healthcare organization in North America was experiencing critical performance degradation across both OLTP transactions and batch processes - impacting finance and HR operations for thousands of users.
The customer was running a complex enterprise stack - Oracle RDBMS, PeopleSoft (Financials and HR), WebLogic web servers, and Tuxedo application servers - and facing performance and availability issues that affected end-users and batch processing across both systems.
They needed four things: performance optimization of specific OLTP transactions and batch processes that were degrading user experience; monitoring capabilities with automation to detect issues before users reported them; alerting features that would immediately notify support teams of component failures; and logging with trend analysis for long-running transactions and processes, enabling proactive capacity planning rather than reactive firefighting.
The environment was complex - multiple infrastructure layers (database, application server, web server, OS) each contributing to performance problems, with no unified view of where bottlenecks actually lived.
Approach
We performed full-stack performance optimization spanning SQL tuning, middleware configuration, OS-level tuning, and built custom monitoring, alerting, and rogue-process detection systems from scratch.
The Periscope Labs engineering team delivered a comprehensive performance optimization and SRE engagement covering seven service areas across four infrastructure layers.
SQL & Database Tuning. Long-running and high-resource-usage queries were identified, prioritized by business impact first, then by database and infrastructure impact. The team performed SQL tuning across both the Financials system and HR system independently, as each had distinct query patterns and bottleneck characteristics. For every tuned query, a detailed document was provided with the exact changes made, an explanation of the tuning rationale, and the full investigation process - creating a knowledge base the client's team could reference going forward.
Middleware & Application Server Tuning. WebLogic web server and Tuxedo application server configurations were reviewed and optimized for connection pooling, thread management, session handling, and resource allocation. These middleware-layer optimizations addressed latency that wasn't visible at the database level alone.
Operating System Tuning. OS-level parameters were reviewed and adjusted to support the optimized database and middleware configurations - ensuring kernel, memory, and I/O settings weren't creating bottlenecks beneath the application stack.
Rogue Runner Detection. A custom "Rogue Runner" system was built to identify and manage runaway processes - queries or batch jobs that exceeded expected duration or resource consumption thresholds. This provided automated detection and notification before rogue processes could cascade into broader system degradation.
Monitoring, Logging & Alerting. Purpose-built monitoring tools were developed or configured for each infrastructure layer:
- Tuxedo Application Server monitoring - real-time visibility into app server health, queue depths, and service response times
- WebLogic Web Server monitoring - connection pool status, thread usage, and request processing metrics
- Operating System monitoring - CPU, memory, disk I/O, and network utilization tracking
- Alerting with immediate email notification of support teams when component issues are detected
- Trend analysis for short-to-long-term review of application performance, enabling proactive configuration and sizing changes
Load Testing. Performance benchmarks were established through controlled load testing, validating that tuning changes delivered measurable improvements under realistic traffic patterns.
Architecture Assessment. The overall system architecture was reviewed to identify structural bottlenecks and recommend longer-term improvements beyond the immediate tuning work.
Outcome
Significantly improved application performance and reliability - with SQL bottlenecks resolved, real-time monitoring across all infrastructure layers, and proactive alerting that catches issues before users do.
The engagement delivered measurable improvements across all four infrastructure layers.
SQL tuning resolved critical bottleneck queries across both Financials and HR systems, with each optimization documented for the client's internal team to maintain and extend. Middleware and OS tuning eliminated latency sources that were invisible at the application layer but contributing to overall response time degradation.
The Rogue Runner detection system provides automated identification of runaway processes, preventing cascade failures that previously required manual intervention. The monitoring and alerting infrastructure delivers immediate notification of component issues across Tuxedo, WebLogic, and OS layers - transforming the operations model from reactive (users report problems) to proactive (systems detect and alert before user impact).
Trend analysis tools enable the client's infrastructure team to review performance patterns over time and make proactive capacity planning and configuration decisions - anticipating problems before they affect production users.
Every tuning change was documented with full rationale, creating an institutional knowledge base that outlasts the engagement.