Analyzing and Optimizing Business System Performance: A Comprehensive Guide

Today we discuss the analysis and diagnosis of performance issues in business systems after they go live, focusing on how to identify and optimize the root causes.

System Performance Issue Analysis Process

When a system shows no performance problems before launch but encounters serious issues after going live, the main scenarios are high concurrent access, data volume growth, and changes in critical environment factors such as network bandwidth.

First, determine whether the problem exists under single‑user (non‑concurrent) conditions or only under load. Single‑user issues are usually easier to test and often stem from code or SQL inefficiencies, while concurrent issues require pressure testing in a test environment.

During load testing, monitor CPU, memory, and JVM to detect problems like memory leaks; even under concurrency, performance anomalies may be caused by the application code itself.

Factors Influencing Performance Issues

Performance factors can be divided into three main areas: hardware environment, software runtime environment, and the software program itself.

Hardware Environment

Hardware includes compute, storage, and network resources. Server compute capability is often described by TPMC, but real‑world performance can differ. Storage I/O performance is a common bottleneck; tools such as iostat , ps , sar , top , and vmstat help monitor CPU, memory, JVM, and disk I/O.

Runtime Environment – Database and Application Middleware

Database and middleware tuning are frequent sources of performance problems.

Database Performance Tuning

For Oracle, performance is affected by system, database, and network factors. Optimization includes disk I/O, rollback segments, redo logs, SGA, and object design. Monitoring can be enabled by setting TIMED_STATISTICS=TRUE in init.ora and ALTER SESSION SET STATISTICS=TRUE , then running utlbstat.sql and utlestat.sql to generate a report.

Application Middleware Performance Analysis and Tuning

Middleware containers (WebLogic, Tomcat, etc.) require configuration tuning, JVM parameter optimization, thread‑pool settings, and connection pool sizing. JVM parameters such as -Xmx , -Xms , -XX:MaxNewSize , -XX:NewSize , -XX:MaxPermSize , -XX:PermSize , and -Xss should be set according to workload characteristics.

Note that newer JVMs replace PermSize with Metaspace, so heap and Metaspace sizing must be balanced.

Software Program Performance Issues

Many performance problems stem from code defects such as large object initialization inside loops, unreleased resources causing memory leaks, lack of caching, long‑running transactions, and sub‑optimal data structures or algorithms. Code reviews and static analysis are essential to uncover these issues.

Extended Considerations for Business System Performance

Pre‑deployment performance testing often fails to fully simulate production hardware, data volume, and concurrency, making post‑launch issues common.

Horizontal scaling of databases and middleware can mitigate some load problems, but single‑node performance must still be optimized when baseline latency is high.

Classification of Performance Diagnosis

Static classification includes operating system/storage, middleware (database and application servers), and software layers (SQL, business logic, front‑end). Dynamic analysis tracks request flow through code and infrastructure to pinpoint bottlenecks.

APM (Application Performance Management) tools provide end‑to‑end monitoring, linking resource usage to specific services, SQL statements, and business functions, greatly improving diagnosis efficiency.

Overall, a systematic approach—monitoring, pressure testing, code review, and targeted tuning of hardware, database, middleware, and application code—is required to resolve business system performance issues.