Guide · DevOps & SRE
Reliability targets for systems that run in the field
Service levels for field systems should reflect what a branch or school actually experiences.
In short: Reliability targets for field systems should be defined from the user's side, such as whether a branch could complete a transaction, not from server dashboards. Use an error budget to balance change against stability, and review every significant outage without blame to find and fix causes.
Systems that run in branches, schools, stations and vehicles depend on devices, local networks, mobile connectivity and central applications. A server can report 99.9% availability while a branch cannot work at all.
Define availability from the user's side
Start with what the user needs to do: open an account, take attendance, record a delivery. Measure whether that task succeeded, from the location where it happens. Synthetic tests from representative sites, and success rates from the application itself, give a truer picture than server uptime.
Set realistic targets by site type
A metro office with fibre and a rural school on mobile data will not achieve the same results. Set targets that reflect each environment, and invest where gaps are largest.
Error budgets for planned change
- A target of 99.5% success leaves a small "budget" of acceptable failures each month.
- When the budget is healthy, teams can release changes more freely.
- When it is nearly spent, they slow down and focus on stability.
This turns arguments about speed versus stability into a shared, data-based decision.
Blameless reviews after every outage
After a significant incident, document what happened, how it was detected, its impact and why it occurred, focusing on systems and processes rather than individuals. Agree actions, assign owners and follow them up. Over time, these reviews remove the causes of repeated outages.
Frequently asked questions
What is a service level objective (SLO)?
A target for how reliably a service should perform, such as the share of successful transactions over a month, measured from the user's point of view.
What is an error budget?
The amount of failure allowed within a reliability target over a period, used to decide how much change a team can safely make.
What is a blameless post-incident review?
A review of an outage that focuses on causes in systems and processes, not on blaming people, so lessons are shared openly and fixed.