SRE - The Magic of Error Budgets - Turning Failure into Strategy
One of the usual scenario I have seen with a product team is releasing new features at lightning speed. While that was great for innovation, the incident volume might started rising, & so did customer complaints.
Although they made sure to have dashboards, alerts, & even SLIs/SLOs in place. Still they might miss something crucial: A way to balance velocity & reliability.
That’s when Error Budgets are introduced, & everything changed.
An Error Budget is the amount of unreliability you’re allowed in a system without breaking your SLO.
Let’s say your SLO is 99.9% availability per month. That means: You’re allowed to fail for 0.1% of the time and that's 43 minutes & 12 seconds of downtime per month.
That "allowed failure window" is your error budget given your SLI's caused DownTime in Application is ZERO. Otherwise Error Budget is Available Application DownTime(SLO's Based) - Application Downtime caused by Application SLI's.
And believe it or not, it’s not a bad thing to use it.