Vocabulary for Site Reliability Engineers
Essential SRE vocabulary explained in plain English: SLO, error budget, toil, blameless postmortem, runbook, and more — with usage examples.
Site Reliability Engineering (SRE) has its own distinct vocabulary — a mix of traditional operations terminology, Google-originated concepts, and software engineering language. For non-native English speakers moving into SRE roles or working alongside SRE teams, mastering this vocabulary is essential for both daily communication and career growth.
This guide covers the core SRE vocabulary with clear definitions, usage examples, and notes on how each term is used in practice.
Service Level Concepts
SLI (Service Level Indicator)
An SLI is a specific metric that measures an aspect of a service’s reliability — for example, the proportion of successful HTTP requests, or the fraction of requests served within 200ms.
“Our primary SLI for the checkout service is the success rate of payment requests.”
“We track three SLIs: availability, latency, and error rate.”
SLO (Service Level Objective)
An SLO is the target value or range for an SLI — the reliability goal your team commits to.
“Our SLO is 99.9% availability, measured over a rolling 30-day window.”
“We missed our SLO for the third consecutive month, which has triggered a reliability review.”
SLA (Service Level Agreement)
An SLA is a formal contract with a customer that defines the expected service level and the consequences of missing it. Unlike an SLO (internal target), an SLA has legal and financial implications.
“Our enterprise customers have an SLA that guarantees 99.95% uptime. Breaching it triggers a service credit.”
Key distinction to remember:
- SLI = what you measure
- SLO = what you target
- SLA = what you promise to customers
Error Budgets
Error Budget
An error budget is the amount of downtime or errors you are permitted to have while still meeting your SLO. If your SLO is 99.9% availability, your error budget is 0.1% — about 43 minutes of downtime per month.
“We’ve consumed 60% of our error budget this month — we should slow down risky deployments.”
“The error budget gives us a data-driven way to decide how much risk to take with new releases.”
Error Budget Burn Rate
The burn rate describes how quickly you are consuming your error budget relative to the expected pace.
“We’re burning our error budget at three times the expected rate — if this continues, we’ll exhaust it in ten days.”
Operational Concepts
Toil
Toil is manual, repetitive, automatable operational work that does not add lasting value. Reducing toil is a core SRE principle.
“Manually restarting services after every deployment is pure toil — we need to automate this.”
“The SRE team tracks toil hours each quarter and aims to keep it below 50% of engineering time.”
Runbook
A runbook is a documented procedure for handling a specific operational scenario — typically an incident or routine task.
“When the payment service latency exceeds 2 seconds, follow the runbook in Confluence to diagnose and resolve it.”
“We’re automating parts of the runbook so on-call engineers spend less time on repetitive diagnostics.”
Playbook
A playbook is similar to a runbook but typically covers a broader range of scenarios or a higher-level incident response strategy.
“The security incident playbook defines the escalation path and communication protocol for data breaches.”
Incidents and Reviews
Blameless Postmortem
A blameless postmortem is a structured review of an incident that focuses on systemic causes rather than individual mistakes. The goal is to learn and improve, not to assign blame.
“After the outage, we ran a blameless postmortem and identified three process gaps that contributed to the incident.”
“Blameless culture is essential — people won’t report near-misses if they fear punishment.”
Mean Time to Recovery (MTTR)
MTTR is the average time it takes to restore service after an incident.
“Our MTTR for P1 incidents is currently 47 minutes — we’re targeting sub-30 minutes by end of quarter.”
Mean Time Between Failures (MTBF)
MTBF is the average time between incidents. A higher MTBF indicates greater reliability.
“Increasing MTBF requires eliminating recurring failure patterns, not just faster recovery.”
Architecture and Capacity
Chaos Engineering
Chaos engineering is the practice of deliberately introducing failures into a system to test its resilience.
“We run chaos experiments in staging to verify that our circuit breakers behave correctly under failure conditions.”
Capacity Planning
Capacity planning is forecasting future resource needs based on growth projections and current usage.
“Based on our traffic growth rate, we’ll need to double database capacity by Q4 — capacity planning should start now.”
On-call
Being on-call means being available outside normal hours to respond to incidents.
“I’m on-call this week — my response SLA for P1 incidents is 15 minutes.”
Key Vocabulary Reference
| Term | Definition |
|---|---|
| SLI | The metric you measure (e.g., success rate) |
| SLO | The target for that metric (e.g., 99.9%) |
| SLA | The customer-facing commitment |
| Error budget | Allowable downtime within the SLO |
| Toil | Manual, repetitive, automatable work |
| Runbook | Step-by-step procedure for a specific scenario |
| Blameless postmortem | Incident review focused on systemic causes |
| MTTR | Average time to recover from an incident |
SRE vocabulary is precise by design — these terms carry specific meanings that enable exact communication about reliability. Learning to use them accurately will help you contribute to SRE discussions, write clearer incident reports, and work more effectively with reliability-focused teams.
Navigating Nuances: A Focus on Clarity for Non-Native Speakers
The core concepts of Site Reliability Engineering – Service Level Objectives (SLOs), error budgets, toil reduction – are often expressed in a very specific, almost technical English. It’s understandable; the precision is vital when dealing with system stability and performance. However, for developers whose first language isn’t English, this can feel overwhelming, leading to misunderstandings or hesitancy in contributing effectively. This section focuses on bridging that gap by highlighting common phrasing variations and offering practical strategies for interpreting and using SRE terminology confidently.
One key area is understanding the difference between descriptive and prescriptive language. Initially, you might hear phrases like “We need to reduce toil” – it sounds directive. But a more helpful framing might be, “Let’s explore ways to minimize repetitive manual tasks that detract from proactive system improvements.” Similarly, stating an SLO isn’t just about numbers; it’s about the user experience. Saying “99.9% uptime” is technically accurate, but explaining why that uptime matters – “This translates to approximately 43 minutes of downtime per month, which directly impacts our users and their ability to complete critical tasks” – provides crucial context. Another frequent area for confusion is around blame. The concept of a “blameless postmortem” isn’t about assigning fault; it’s about learning from incidents without judgment. Phrases like “What can we do differently next time?” are far more productive than accusatory statements.
Furthermore, pay attention to the subtle differences in how urgency is conveyed. The phrase “urgent” often carries a strong emotional weight. Instead of saying something is “urgent,” try phrasing it as “requiring immediate attention” or “impacts critical operations.” This avoids unnecessary panic and allows for a more rational assessment of the situation. It’s also crucial to recognize that direct translations of technical terms can be misleading. For example, “scaling” doesn’t always imply rapid growth; it simply means adapting to changing demands.
Finally, don’t hesitate to ask for clarification. SRE teams generally appreciate proactive learners who are genuinely interested in understanding the underlying principles. A simple question like, “Could you elaborate on what ‘capacity planning’ entails in this context?” demonstrates engagement and a willingness to learn – qualities highly valued within an SRE environment.
# Example: Monitoring CPU usage with Prometheus and Grafana
# This is a common scenario discussed when reviewing performance metrics.
grafana_query='{
"query": "sum(rate(node_cpu_seconds_total{mode!="idle"}[5m])) by (instance)",
"title": "CPU Usage - Node Instance"
}'