Skip to main content
Mental Models

Redundancy: Margin of Safety

Redundancy involves having duplicate resources (like backups or fail-safes) to prevent failure when something goes wrong, providing a margin of safety.

THE IDEA

REDUNDANCY: MARGIN OF SAFETY

Redundancy is ambiguous because it seems like a waste if nothing unusual happens. Except that something unusual happens—usually.

—Nassim Taleb

Redundancy gives a system duplicate resources or alternative ways to perform a function if one component fails. A backup connection or power supply can keep a failure from disabling the whole system.

For example, having redundancy in our Internet connection means having two connections available so that in case one fails we can still use the other.

Having redundancy or duplication of power supply is common in other areas outside of computing, for example in hospitals. In a hospital, an operating room cannot be without electricity (the life of the patient is at stake) so there are generators in place that run on diesel and start as soon as the electricity goes out.

In aviation, critical flight instruments and communications systems may be tripled in the cockpit. Structural load capacity provides a different kind of safety margin: the source gives wing-load figures of around three times the aircraft’s weight (3G) and a breaking load of 4.5 G. Those figures concern structural capacity rather than duplicated components.

In computer science, the source distinguishes four forms of redundancy:

  • Hardware or physical redundancy: This type of redundancy consists of incorporating additional hardware, usually in order to detect faults or achieve a high tolerance for them. It is the most frequently used technique.
  • Software redundancy: This type of redundancy consists in adding additional lines of code to the programs to avoid errors, detecting them or correcting them. This supplementary code can have different functions, such as preventing data from getting out of range, preventing arithmetic errors …
  • Information redundancy: This type of redundancy consists of adding additional information to the data to allow the detection or correction of errors. Here we could include methods like parity codes or checksums.
  • Time redundancy: This technique consists of using a small software that repeats the calculations several times to compare the results and see if there is any difference.

When to use it

  • Ensuring uninterrupted internet for remote work by maintaining a secondary connection.

  • Designing a data center with N+1 redundant power supplies and cooling systems.

  • Adding error-correcting codes to communication protocols to detect and fix data corruption.

  • Performing critical calculations multiple times with slight variations to validate results (time redundancy).

FROM THEORY TO EVERYDAY LIFE

In practice

01

Internet Redundancy

Having two Internet connections available so that if one fails, the other keeps you connected without interruption.

02

Hospital Backup Power

Operating rooms use backup power systems, including generators that start automatically during an outage, to help maintain essential equipment.

03

Aviation Safety Margins

Critical flight instruments may be tripled in a cockpit. Separately, the source illustrates structural safety margins with wing-load figures of about three times the aircraft’s weight (3G) and a breaking load of 4.5G. Structural capacity and duplicate instruments address different failure risks.

04

Computer Science Redundancy Forms

Hardware redundancy adds duplicate physical components; software redundancy adds extra code for error detection/correction; information redundancy uses parity codes or checksums to protect data; time redundancy repeats calculations and compares results to detect errors.

✳

A MOMENT TO REFLECT

Put it to work.

  1. Where in your life or work could a backup system prevent a critical failure?

  2. How do you balance the cost of adding redundancy against the potential cost of a single point of failure?

  3. Which of the four types of redundancy (hardware, software, information, time) could you apply to improve reliability in a current project?

CONTINUE THE CONVERSATION

Source texts

  • Wisdom Theory 100 Mental Models (2020) (Wisdom Theory)
← More mental models