When performance starts to degrade, most teams follow the same instinct: add more instances, bring in GPUs, scale resources…and watch costs grow faster than actual performance improvements.

In this session, we’re going to challenge that default reaction and replace it with something far more effective: a data-driven way of understanding what your systems are actually doing in production. We’ll explore why CPU utilization (one of the most commonly used metrics) is often misleading, how to identify the true shape of a workload, and why a significant portion of production systems can run more efficiently (and more cost-effectively) on modern CPUs than most teams expect.

Through examples and workload analysis, you’ll see how to move from “just add more hardware” to “understand before you scale,” using evidence instead of intuition. The goal is simple: help you design systems that balance performance, efficiency, and cost from day one…not as an afterthought.

You’ll walk away with:

A clear mental model to understand your workload behavior
A practical, repeatable framework you can apply immediately to decide when scaling is necessary…and when it’s just unnecessary spend.

Because optimization isn’t about adding more…it’s about understanding better.