3.0 - Energy Efficiency¶

3.1 - Introduction¶

The energy efficiency principle is simple: Use the least amount of energy possible
Energy = Ability to do work
Multiple forms of energy exist and can each be converted to another in some form e.g. chemical to electrical conversion happens during coal burning.
In the example above, electricity is the secondary energy converted from another type; this allows measurement of electricity usage.
All software consumes electricity, the best way to reduce consumption and subseqeuent carbon emissions is to make applications more efficient.
Additionally, green software practitioners take responsibility for the energy consumed by their products, and design them to consume as little as possible by default.
At every step of the process, the waste between steps should be minimised and efficiency should be maximised.
The final area of consideration is the user, how can, when they're running the application, minimise their unnecessary emissions.
This is managed by how the application is designed, such as batching jobs together.

Most electricity is produced via fossil fuel burning.
The fuels are made from decomposing plants and animals, found within the earth's crust, and comprise of carbon and hyrdrogen.
Typical examples: coal, oil, natural gas.
As most electricity comes from fossil fuels, the most significant carbon emission cause, one can immediately see how tightly linked electricity usage is to fossil fuel emissions.
To be carbon efficient, one therefore needs to be energy efficient as energy is a proxy for carbon.

Clean energy comes from renewable, zero-emission sources that don't pollute the atmosphere when used.
They also save enrgy through energy efficient practices.
There are often overlaps between clean, green, and renewable energy sources, key differences include:
Clean Energy - Doesn't produce carbon emissions e.g. nuclear
Green energy - relies on sources from nature
Renewable Energy - sources don't expire e.g. solar, wind

Energy is measured in Joules (J); the SI unit of energy.
Power is measured in watts (W), 1W = 1 Joule per second
Kilowatt (kW) = 1000 Joules per second
kilowatt-hour (kWh) = measure of energy corresponding to 1 kilowatt sustained for 1 hour.

Data centers use Power Usage Effectiveness (PUE) to measure data center energy efficiency.
This defines how much energy the infrastructure uses compared to cooling and other supporting overheads.
Examples:
1.0 - Computing is using nearly all the enrgy
2.0 - An additional watt of IT power is needed to cool and distribute power to the infrastructure for every watt it uses.
Effectively, PUE = a multiplier to an application's energy consumption.
Example: if an application uses 10kWh and the data center's PUE is 1.5, the actual consumption is 15kWh; 5kWh goes towards operational overhead of the data center and 10kWh is used to run the application.

Measures the relationship between power consumed by a computer and the rate at which useful work is done (its utilization).
Utilization = how much of a computer's resources are used, usually expressed as a percentage.
Power usage and utilization isn't proportional or linear in relationship, plateuing at high power usage.
The more a computer is utilized, the more efficient it becomes at converting electricity to computation operations.
To improve hardware efficiency, workloads should be run on as few servers as possible; this would maximise utilization rate and therefore energy efficiency.

How much electricity is drawn when in an idle state.
Typically varies by configuration and hardware components, but all parts have it.
It's the reason that lots of devices have power-saving modes; where the screen and disk are put to sleep, or the CPU's frequency is changed.
These modes save electricity, but typically have adverse effects on performance e.g. slow restarts.
Servers aren't typically configured for aggresive or minimal power saving.
Typical use cases for servers demand total capacity as quickly as possible, as the server needs to respond to rapidly changing demands.
This leads to many servers in idle modes during low demand periods; which incurs carbon costs from both embedded carbon and inefficient utilization.

Electricity is a proxy for carbon, so building an application that is energy efficient is equivalent to building an application that is carbon efficient.
Green software takes responsibility for its electricity consumption and is designed to consume as little as possible.
Quantifying the energy consumption of an application is a step in the right direction to start thinking about how an application can operate more efficiently. However, understanding your application's energy consumption is not the only story.
The hardware your software is running on uses some of the electricity for operational overhead. This is called power usage efficiency (PUE) in the cloud space.
The concept of energy proportionality adds another layer of complexity since hardware becomes more efficient at turning electricity into useful operations the more it's used.
Understanding this gives green software practitioners a better insight into how their application behaves with respect to energy consumption in the real world.