System Performance Troubleshooting Guide

CPU

Start with top to find the process that uses the most CPU.

Use flame graphs to see which operations take the most time. Then optimize those specific operations.

Watch for frequent system calls.

Multiple threads can reduce execution time. Heavy traffic might overload your CPU. Consider using LVS for load balancing.

Slow queries can also cause problems. Full table scans that compare each row consume CPU resources.

Memory

Use top to check memory usage.

This often happens when code allocates memory frequently. Languages with garbage collection increase GC time. This affects CPU indirectly.

You can optimize by using lazy allocation strategies. Trade time for space. Reuse objects instead of creating new ones.

Disk I/O

Database issues cause most disk I/O problems. Slow query logs help locate problematic statements.

Don’t assume log writes are fast. Under heavy traffic, logging triggers multiple read operations. This slows down all I/O.

Long write operations benefit from larger database write buffers. More data goes to memory first, then flushes to disk later.

Lazy writing: Control write frequency. Keep data in memory until it reaches a certain amount, then write to disk every few seconds.

Consider upgrading your hard drives.