Tech Tip #2: Your cloud servers are mostly idle

Most cloud servers sit idle about seventy percent of the time.

Video created by explainapaper.com

The problem

Published studies put cloud servers active only 10 to 30% of the time, yet an idle machine still draws a significant fraction of its peak power. Servers are bought for the busiest moment, and slow software makes that moment bigger than it needs to be.

The fix you can use today

  • First cut the load: fix the wasteful code so each request needs less.
  • Then right-size: fewer or smaller machines for what is left.
  • Let quiet services scale to zero, and look for machines idle all day.

Studies reviewed by GreenCode researchers found that many cloud servers are busy only 10 to 30% of the time. GreenCode's research review draws on this work.

Tools that help

  • GreenCode Constitution skill (GreenCode researchers): open-source energy principles and a catalogue of 37 energy anti-patterns that an AI coding agent can follow to find and fix waste like this
  • Scaphandre: measures the power drawn by each process on a Linux server and exports it to Prometheus and other monitoring stacks
  • Kepler: exports energy estimates per pod and container on Kubernetes, using eBPF and hardware counters

How GreenCode helps

GreenCode helps tackle this by mapping a running estate and tracing its energy and CO2e to the services that cause it. That shows which services run hot and which are barely running at all. More on this.

Tags

#idle#right-sizing#cloud#cost

Sources

Transcript

Most cloud servers sit idle about seventy percent of the time. They are busy only ten to thirty percent of the time. That comes from studies reviewed by GreenCode researchers. GreenCode's research review draws on this work. But idle machines still draw a big share of their full power. Servers are bought for the busiest moment, and slow software makes that moment bigger. So you pay for electricity, cooling and hardware that mostly wait. Here is the fix you can use today, in the right order. First, cut the load, by fixing the wasteful code. Then right-size: fewer or smaller machines for what is left. And let quiet services scale down to zero. Check your dashboards for machines that are mostly idle, all day. Kepler shows each Kubernetes pod's energy; Cloud Carbon Footprint estimates yours from cloud bills. Shrink the machines first, and you just size them for the waste. GreenCode helps tackle this by tracing energy and carbon to the services that cause it. That shows which services run hot, and which barely run at all. That was GreenCode Tech Tip number two. Follow for the next tip, and find more on greencode.ai/tech-tips.

All GreenCode Tech Tips · All GreenCode videos

  • Tip #2
  • For Ops and infrastructure
  • Length 1:12
  • Category Cloud and infrastructure
GreenCode

Want this checked on your own system? Talk to the GreenCode team.

Get in touch