CockroachDB: The Definitive Guide

Chapter 15. Cluster Optimization

It doesn’t matter what database, framework, application, or device you’re responsible for—someone always wants it to be faster. This is probably one of the most persistent challenges for computer scientists. One can imagine after Alan Turing created his breakthrough decoding machine during World War II that his superiors said, “Well done Alan, but can it be faster?”

It’s not surprising that performance optimization is such a perennial activity. It’s always been important to provide users with responsive applications and to be able to maintain acceptable throughput for batch processes. However, in the modern era, performance has become even more important. In the internet era, poorly performing customer-facing applications lead to customers abandoning your online services, directly affecting the bottom line. In the cloud era, performance optimization is cost optimization—a poorly performing application will consume more cloud computing resources, which will increase your monthly bill.

Tuning Versus Firefighting

Performance optimization usually occurs in one of two forms:

Performance firefighting

A critical performance problem arises and must be addressed immediately.

Performance tuning

The system is optimized systematically to improve its performance and reduce the cost of operation.

The more performance tuning you do, the less performance firefighting you’ll need to undertake. Nevertheless, for almost all database administrators, you’ll ...