The article explains why engineers usually trade memory for faster response, illustrates the choice with caching and batch‑processing examples, and provides a decision framework based on IO bottlenecks, access frequency, data volatility and resource costs.
An Oracle 10g database froze at 10 am due to a massive log file sync wait, traced through AWR analysis to I/O performance problems and excessive commit activity, with root‑cause verification via LGWR switch frequency, parallel‑write latency, and alert log errors, leading to practical remediation steps.
This article outlines a comprehensive approach to system performance optimization, covering key metrics, profiling tools, common bottlenecks in CPU, memory, disk and network I/O, and practical guidelines for continuous monitoring and improvement in production environments.
The article explains how IO and CPU bottlenecks increase active database connections, then details horizontal and vertical sharding techniques, practical tools, step‑by‑step implementation, and common pitfalls to help engineers relieve pressure on database resources.
High system load can occur even when CPU usage is low, typically due to many processes waiting for disk I/O; this article explains load concepts, process states, scheduling, and common scenarios such as excessive I/O requests, unindexed MySQL queries, and faulty external storage that cause such bottlenecks.
A sudden high‑load alert on a web server was traced to Nginx’s massive write operations on fastcgi_temp files, revealed through top, iotop, lsof, strace and iostat, and temporarily resolved by redirecting fastcgi_temp to memory.
