Redundancy in IT isn’t


I had a shower thought of my own this morning, in the shower. Like a gentleman. Redundancy, as we understand it in this industry, isn’t.

All the online dictionary I’ve read equates redundancy with being superfluous, unnecessary, or surplus to requirements. Collins takes it a step further by also claiming it can mean:

duplicated or added as a precaution against failure, error, etc

This is what I understand redundancy to mean. I want offsite backups, multiple copies of data in a RAID or OpenZFS pool, and read replicas of my live databases. These are all forms of redundancy, because they can improve performace by spreading load, continue to operate in a degraded state and, most critically, recover from loss.

If we consider redundancy to be an essential requirement of a system then, is the existence of said redundant data not redundant? Anyone who claims backups are redundant in the traditional English use of the term isn’t qualified to design or operate computer systems!

Much as my old boss said that anything that isn’t documented doesn’t exist, I’d argue any data without redundancy is ephemeral. If your system can’t tolerate that, redundancy is therefore a requirement.

Each industry has its own vocabulary and nomenclature that may conflict with definitions in other spheres. It’s why nerds get up in arms about the word hacker, despite violent and negative connotations predating our use (Jack the Ripper wasn’t hacking people up to fix them in an innovative way). I wonder what other terms don’t just have different meanings, but are polar opposites?

I always thought OSPF sounded like a laundry detergent.

Author bio and support


Ruben Schade is a technical writer and infrastructure architect in Sydney, Australia who refers to himself in the third person. Hi!

The site is powered by Hugo, FreeBSD, and OpenZFS on OrionVM, everyone’s favourite bespoke cloud infrastructure provider.

If you found this post helpful or entertaining, you can shout me a coffee or send a comment. Thanks ☺️.