Article

Embracing Risk (Error Budgets)

Google SRE Team · Google SRE Book Free Advanced

Google's error budget is the most rigorous version of the one third rule: define how much unreliability you can tolerate, and when you blow past it, new feature work stops until reliability recovers. It gives you an objective trigger for when to drop the exciting build and pay down reliability. Even a rough version of this keeps promises from silently rotting.

• Set an explicit budget for how much unreliability you accept
• When the budget is spent, feature work pauses for reliability
• A shared budget ends the endless features versus stability argument