Site Reliability Engineering (SRE)

Process internals, memory model, file descriptors, networking stack, kernel tuning, and performance analysis with perf, strace, and eBPF.

• The Linux Command Line (free book)docfree
• Brendan Gregg: Systems Performance (book)coursepaid
• Linux Performance Analysis in 60 Seconds (Netflix blog)articlefree

• Diagnose CPU bottleneck with top/perf
• Trace system calls with strace
• Analyse memory leak with valgrind / /proc
• Network tuning: sysctl TCP parameters

SLIs, SLOs, SLAs, error budgets, toil elimination, and the SRE vs DevOps mental model. Chapter-by-chapter reading with applied exercises.

• Google SRE Book (free online)docfree
• Google SRE Workbook (free online)docfree
• School of SRE: Linkedin Engineering (free)coursefree

• Write SLOs for a hypothetical e-commerce service
• Calculate error budget burn rate
• Identify toil in a daily workflow and automate it

Scripting, tooling, and automation: Python for ops scripts, Go for reliable CLI tools and services. Focus on practical SRE use cases.

• Automate the Boring Stuff with Pythondocfree
• Go Tour (free official)coursefree
• Python for DevOps (O'Reilly)coursepaid

• Python: health-check script that pages on failure
• Go: build a simple HTTP load-test tool
• Automate runbook steps as a script

PromQL, recording rules, alerting rules, Grafana dashboards, and the RED and USE methods for service health.

• Prometheus Official Docsdocfree
• Grafana: Getting Starteddocfree
• Robust Perception: PromQL for Humansarticlefree

• Instrument a Node/Python service with client library
• PromQL: p99 latency, error rate, saturation
• Grafana: RED dashboard (Rate, Errors, Duration)
• Alertmanager: page on SLO breach

Structured JSON logging, log aggregation with Loki or Elasticsearch, Logstash pipelines, Kibana/Grafana queries, and log-based alerting.

• Elastic: Getting Started with ELK Stackdocfree
• Grafana Loki Docsdocfree

• Structured log format: trace_id, service, level
• LogQL: find all 5xx errors in last 1h
• Correlation: trace a request across 3 services

Spans, traces, context propagation, sampling strategies, and correlating traces with metrics and logs (the three pillars).

• OpenTelemetry Docsdocfree
• Jaeger Tracing Docsdocfree

• Instrument a service with OTel SDK
• View trace in Jaeger: identify slow span
• Correlate a trace to a Prometheus spike

Incident response lifecycle, postmortem culture, on-call best practices, PagerDuty setup, runbooks, and blameless retrospectives.

• PagerDuty Incident Response Docs (free)docfree
• Google SRE Book: Chapter 14 - Managing Incidentsdocfree
• Increment: On-Call Issuearticlefree

• Write a postmortem for a real or simulated incident
• Create a runbook for top-3 alert types
• Configure PagerDuty escalation policy

Game days, fault injection, Chaos Monkey, Gremlin, blast radius limiting, and recovery testing.

• Chaos Engineering: O'Reilly Book (free preview)docfree
• Netflix TechBlog: Chaos Monkeyarticlefree
• Chaos Toolkit (open source)repofree

• Run a game day: kill a service instance
• Verify graceful degradation under load
• Chaos experiment: latency injection on a dependency

Pod autoscaling (HPA/VPA/KEDA), disruption budgets, priority classes, resource quotas, node affinity, and SRE-focused Kubernetes patterns.

• Kubernetes Docs: Production Best Practicesdocfree
• Learnk8s: Production Kubernetes Checklistarticlefree
• Mumshad Mannambeth: CKA (Udemy)coursepaid

• Configure HPA on a deployment
• PodDisruptionBudget: survive a node drain
• Resource requests/limits: avoid OOMKill
• Pass CKA exam (target)

Apply everything to a real service: define SLOs, instrument metrics + logs + traces, build dashboards, configure alerts, write runbooks, and run a chaos game day.

• Google SRE Workbook: Implementing SLOsdocfree
• awesome-sre: Curated SRE resources (GitHub)repofree

• SLO document approved by stakeholders
• Dashboards covering RED + USE methods
• Alerting with no false positives for 2 weeks
• Game day executed and postmortem written