Google SRE book definition: manual, repetitive, automatable work that scales linearly with service growth and has no enduring value.
Examples:
- Manually rotating credentials every 90 days
- Manually patching servers
- Manually responding to "is the service up?" tickets
- Manually restarting a stuck service every Tuesday
The opposite of toil is engineering work that reduces future toil. A healthy SRE function aims for 50% engineering, 50% operations at most.