What is the difference between slack and waste in production ops?
Waste is when resources are deeply over-provisioned, underutilized, or not utilized at all.
On the other hand, slack appears like the same thing, but you create it with purpose.
Here’s why we create slack:
You can't operate your systems at 100% because you need to deal with:
- variability in demand, such as usage spikes
- background activities that may get prioritized out (you will have to to catch up with them later in foreground latency)
- disk, node, and network failures, where you need to rotate demand to things that are still around
So slack is the excess resources you keep around to deal with unexpected demands, activity loads, and failures.
In upcoming videos, I’ll talk about how I’ve reduced both slack and waste in the past.