SRE Practitioner study plan

Write out each core SRE principle (embracing risk, SLOs as the reliability contract, toil elimination, monitoring distributed systems, automation over heroics) with one concrete example from your own organisation per principle

Classify eight real behaviours or past incidents from your experience into SRE anti-patterns such as alert-driven heroics, change avoidance, and toil-as-a-service

Author two SLI specifications (one availability, one latency) for a single user journey, justifying the measurement point, aggregation, and window

Compute a 28-day error budget from a proposed SLO and state what fraction of the budget a weekly release cadence would consume

Define SLIs for a multi-service request path and justify where to instrument (client, load balancer, service, dependency) at each hop

Derive multi-window multi-burn-rate alerting rules from an SLO and state the false-positive/false-negative trade-off of each window pair

Instrument a single service across metrics, logs, and traces, then demonstrate correlation of one metric spike to a specific trace

Select among request-based, windowed, and resource-based SLI shapes for three distinct workload profiles and defend each choice

Trace one request end-to-end and annotate the waterfall with the span attributes required to answer 'which dependency caused this latency'

Specify a sampling strategy and span-attribute naming convention for one service, and state what observability data is deliberately excluded and why

Enumerate the trust boundaries in one request path and state the authentication and authorisation decision made at each boundary

Write a blast-radius analysis for one compromised component under a zero-trust assumption, listing what the attacker can and cannot reach

Specify a resilience pattern set (timeouts, retries with jitter, circuit breaking, load shedding, bulkheads) and name the failure mode each pattern addresses and each pattern's failure mode

Build a threat model for one service with at least six threats and a matching mitigation, detection signal, and residual risk for each

Estimate the cost of toil for one team in hours per week and propose a concrete toil budget with a target reduction percentage

Draft an SRE adoption model (consulting, embedded, ownership) with entry and exit criteria for each stage and the organisational incentives that make it stick

Define the schema and join keys for an incident telemetry pipeline that unifies alerts, deploy events, traces, and on-call actions into one queryable record

Write four anomaly-detection or predictive-alerting hypotheses and state the ground-truth labels and evaluation metric needed to validate each before shipping it

Assign incident roles (incident commander, communications lead, operations lead, scribe) and write a responsibility matrix showing who decides what

Draft a four-level severity matrix with an explicit escalation trigger and notification scope for each level

Run a tabletop exercise for a Sev-1 affecting a shared distributed dependency and record time-to-detect, time-to-decide, and time-to-mitigate

Write customer-facing communication templates (initial, update cadence, resolution) for three severity levels with a stated cadence rule

Define a platform golden path for one service type, naming each paved-road component and its self-service interface and owner

Select four platform adoption metrics (time to first deploy, percentage of services on the golden path, change failure rate, lead time) and capture their current baseline values

Write five falsifiable steady-state hypotheses for one system, each with a measurable indicator and a threshold that defines normal

Design three chaos experiments mapped to specific failure modes, each with scope, injected fault, blast radius, and abort condition

Define a stop/abort procedure with explicit numeric thresholds and a named decision owner for unattended experiment runs

Plan a game-day agenda with roles, timeboxes, safety controls, and a rollback path for every experiment

Execute or dry-run one chaos experiment end to end and record whether the steady-state hypothesis held, with the evidence that proves it

Rank all nine domains by mock-exam accuracy and close the five largest gaps with targeted review notes

Re-derive error-budget arithmetic and burn-rate alert thresholds from memory on a blank page without reference material

Complete a timed 60-question mock exam and itemise every miss by domain and root cause (knowledge gap, misread, or unknown term)