INFRASTRUCTURE LEARNINGINFRASTRUCTURE LEARNING
インフラ大全インフラ大全IT INFRASTRUCTURE / LEARNING
Infrastructure learning guide
Find a lesson or lab/

Esc to close · Search terms stay on your device

LEARNING日本語

Topic lessonTopic: Audit coverage and recovery design

AWS Regional outage recovery: keeping a single writer

Choose which Region can accept writes before moving traffic. Follow fictional recovery evidence to decide when to proceed, hold or abort.

What you will learn

Order incident confirmation, write isolation, standby checks and gradual resumption using observed evidence and assigned roles.

Prerequisite

This is a tabletop exercise for a fictional two-site application. It contains no live fault injection or service-specific promotion commands.

Check the business impact and the alternative route first

AWS recovery guidance calls for procedures independent of the failed site and a way to halt unsafe recovery. This scenario allows reads and writes to resume independently.

Evidence that permits each stage of recovery
Evidence that permits each stage of recovery1: Confirm impact — Incident lead checks impact — Isolate East. 2: Isolate East — Recovery lead verifies — Check West. 3: Check West — Business owner authorizes — Resume123Confirm impactIsolate EastCheck WestResumeEvidence that permits each stage of recovery1: Confirm impact — Incident lead checks impact — Isolate East. 2: Isolate East — Recovery lead verifies — Check West. 3: Check West — Business owner authorizes — Resume123Confirm impactIsolate EastCheck WestResume
  1. Confirm impact → Isolate EastIncident lead checks impact
  2. Isolate East → Check WestRecovery lead verifies
  3. Check West → ResumeBusiness owner authorizes
Confirm impact
East order writes are unstable
Isolate East
Stop old-side updates
Check West
Verify consistent data and dependencies
Resume
Only West accepts writes

Hold updates while isolation is uncertain

Fictional application with order IDs shared across the two sites
TimeObservationDecision
10:00East order writes failDeclare the affected scope
10:04West reads work; East write status is unknownHold West writes
10:09East writes stopped; West recovery point confirmedRun a synthetic West order
10:13Synthetic order and read-back succeedResume on a limited route

Recovery automation needs a clear trigger and a way to stop. An unknown state can hide an active write path, so starting writes on the new side could permit both sides to update.

Define what to do when evidence is missing

Separate the sources of evidence. Plan how to confirm state from the standby site when control in the failed site is unavailable. Verify that approach in a drill.

Trace uncertain writes. Distinguish accepted, committed and unresolved operations. Decide whether to replay only after deduplication and reconciliation checks.

Treat failback as a new switch. After the old site recovers, recheck data differences and write authority before directing traffic back.

Check your decision

Choose an option to see the result and explanation.

Why hold writes at 10:04?
Read the answer and reason

Answer: A single writer has not been confirmed · East may still be accepting writes.

What is the next step at 10:09?
Read the answer and reason

Answer: Run a scoped synthetic order · After isolation and recovery-point checks, run a bounded business validation.

East monitoring turns healthy again. What should happen?
Read the answer and reason

Answer: Authorize and reconcile a separate switch · Failback requires a fresh decision on data and write authority.

Check, then explain in your own words

  1. Record impact and current write status.
  2. Check isolation and standby dependencies.
  3. Verify a synthetic transaction before gradually reopening.
Check your understanding: Is a successful read in West enough to enable writes there?

Wait for evidence that East writes have stopped and West has a usable recovery point.

Read access and safe write authority are separate facts.

Sources and conditions checked

The official references used in this lesson are collected here with their scope, conditions and corresponding sections. Independent expert review and live AWS comparison have not been performed.