Planning meeting — Product asks for a '100% uptime' target for checkout in the new customer contract
Product wants 100%
You're the SRE in the room with the product manager and the engineering lead.
Briefing
A large B2B customer wants an availability commitment for ShopLite's checkout API. Priya (product) suggests 100%, 'because that's what customers want'. Current measured availability over the last quarter is 99.83%.
Your shift
Decide before you look. Each choice is locked once made; the next decision unlocks after it. There's no perfect path in real incidents either, only better and worse judgement under uncertainty.
Priya: 'Can we just commit to 100%? Anything less sounds bad.'
How do you respond?
Debrief
Good SLOs start from user experience (an SLI measured close to users, covering errors and latency), are set from what users need and what the system can achieve, sit above any contractual SLA, and are reviewed regularly. 100% is never the answer, because every nine costs roughly ten times more and forbids change.
The postmortem
Blameless postmortem · draft
Summary. Planning outcome: checkout availability SLO defined and agreed with Product.
Impact. N/A — planning session.
Timeline
- 10:00Request for a 100% commitment
- 10:20SLI agreed: non-5xx within 1 s at the load balancer
- 10:40SLO 99.9% / 28d; SLA 99.5% monthly
Root cause. N/A
Contributing factors
- Previously no written SLO, so reliability targets were set by opinion.
Action items
- processPublish the SLO document (SLI query, target, window, owner, budget policy)sre
- detectBurn-rate alerts on the new SLO (Observability course, Case 4.3)sre
- processQuarterly SLO review with Productpriya
The practice behind it
Each nine costs about ten times more
Going from 99% to 99.9% might mean better deploys and monitoring. 99.99% usually needs multi-AZ everything, automated failover, and careful change management. 99.999% (5 minutes a year) typically needs multi-region active-active systems and a dedicated team. Choose the target that the product's users actually need; spend the difference on features.
Your turn
A browse-only product page and the checkout API: should they have the same SLO? Why or why not?
Interview questions
How would you choose an SLO for a new service?