Resilience Bundle

AWS Backup (including the newer Logically Air-Gapped Vault), AWS Vault Lock, and cross-account/cross-region disaster recovery patterns via AWS Elastic Disaster Recovery (DRS). On the Security Specialty exam, resilience isn't really about uptime — it's about whether your recovery data and recovery PATH survive the same incident that took down production, especially ransomware and a fully compromised account.

Data Protection — primary role Governance — immutability/retention Responsive — recovery execution
Immutable
Logically air-gapped vault, by default
Nov 2025
KMS CMK support added to LAG vaults
72 hrs
Vault Lock compliance-mode grace period
MPA
Multi-party approval for recovery

The Core Mental Model — Recovery Has to Survive the Incident

The Question Behind Every Resilience Scenario
⚠️ The Recurring Exam Theme

Resilience questions on this exam are really asking: "if the production account/credentials are fully compromised, does your backup survive?" A standard backup vault in the SAME account as production — even with Vault Lock — is still reachable by a sufficiently privileged compromised identity in that account (Vault Lock protects against accidental/malicious deletion attempts subject to IAM, but doesn't change which account the data lives in). The textbook answer to "ransomware-resilient, cross-account-isolated backup" is almost always a Logically Air-Gapped Vault shared to a separate recovery account via RAM — not just Vault Lock alone.

Exam Domain Mapping

DomainWhere This Bundle Shows Up
Data ProtectionThe centerpiece — backup immutability, encryption of recovery points, retention enforcement
Threat Detection & Incident ResponseRansomware recovery scenarios, GuardDuty malware scanning of backups before restore
Management & Security GovernanceCross-account backup sharing via RAM, Multi-party approval governance, Backup Audit Manager compliance controls
Identity & Access ManagementCross-account DRS failback IAM roles, recovery-account access governance

Decision Tree — Mental Model

Threat

Ransomware encrypting/deleting production data AND any backups reachable from the compromised account; full account compromise eliminating the owning account's ability to authorize its own recovery; AZ/Region outage

Security Goal

Ensure recovery data is immutable, isolated from the production account's own blast radius, and recoverable even if the production account itself is inaccessible

AWS Service

The Resilience Bundle

AWS Backup (standard vaults) Vault Lock (WORM immutability) Logically Air-Gapped Vault (cross-account isolation) AWS DRS (continuous replication + failover)
Implementation

Centralize backup policy via Organizations backup policies; copy critical backups to a LAG vault in a separate recovery account; enable Multi-party approval for that vault; replicate critical EC2 workloads via DRS to a separate recovery Region/account.

Monitoring

GuardDuty malware scanning of recovery points before restore; Backup Audit Manager controls for cross-Region/cross-account copy and Vault Lock compliance; failover/failback drill results.

Remediation

Direct restore from the LAG vault in the isolated recovery account; DRS failover to recovery instances; structured failback (reversed replication) once the primary is clean and restored.

Final Summary

Must Memorize
  • Logically Air-Gapped Vaults are immutable BY DEFAULT (auto-compliance-lock) — no separate Vault Lock configuration step required
  • Vault Lock Compliance mode = truly immutable after the grace period (~72 hrs) — not even root can shorten retention or delete while recovery points exist; Governance mode is removable by authorized IAM principals
  • LAG vaults share via AWS RAM to a separate recovery account for direct restore — this cross-account isolation is the key ransomware-resilience property
  • KMS customer-managed key support for LAG vaults was added November 2025 (previously AWS-owned key only)
  • DRS supports both cross-Region AND cross-account failover/failback, requiring pre-created Failback/right-sizing IAM roles in both directions
Must Understand
  • Why a standard backup vault in the SAME account as production doesn't fully solve the "compromised admin credential" threat model
  • Multi-party approval's role: authorizing recovery access even when the vault-OWNING account itself becomes inaccessible
  • The distinction triangle: standard vault (baseline) vs Vault Lock (immutability within the account) vs LAG vault (immutability + cross-account isolation)
Can De-prioritize
  • Exact dollar pricing per GB stored/restored
  • Console UI navigation specifics
  • Full list of every AWS service AWS Backup supports

Exam appearance probability: MEDIUM-HIGH

Core Services — Deep Dive

AWS Backup — Core Concepts Centralized, policy-driven
What it doesCentralizes and automates backup across AWS services (EC2, EBS, RDS, DynamoDB, EFS, FSx, S3, EKS, and more) under unified backup plans, retention rules, and lifecycle policies
Backup vaultThe logical container storing recovery points — by default a STANDARD vault, reachable/deletable by sufficiently privileged IAM principals in the same account
Organizations backup policiesCentrally enforce consistent backup plans across every account in an AWS Organization, ensuring no account "forgets" to back up a resource type
Cross-Region / cross-account copyA backup plan can automatically copy recovery points to a different Region and/or a different account as part of the same backup job
  • GuardDuty integration: malware scanning of backups can occur BEFORE a restore is completed, catching ransomware/malware that may have been silently captured into the backup itself.
AWS Backup Vault Lock WORM immutability
Governance modeProvides WORM protection, but IAM principals with specific permissions CAN still remove the lock or modify retention settings — flexible, but not truly tamper-proof against an insider/compromised admin
Compliance modeOnce the grace period (typically 72 hours) expires, the lock becomes PERMANENT — the vault cannot be deleted while recovery points exist, and the retention period cannot be shortened by ANYONE, including the root user
  • Vault Lock protects against accidental or malicious DELETION/RETENTION-SHORTENING within the rules of the lock — it does NOT, by itself, move the data out of the production account's blast radius. A compromised account with sufficient access could, in theory, still reach a Vault-Locked vault in that same account (though not delete protected recovery points in Compliance mode) — this is precisely the gap LAG vaults close with cross-account isolation.
  • "Enforce a retention period that even our own root user cannot shorten, for regulatory compliance" → Vault Lock in Compliance mode.
Logically Air-Gapped (LAG) Vault High — the sharpest 2024-2026 addition
StatusGA since August 2024 (preview since 2023); a continuously expanding-coverage feature
Default protectionImmutable backup copies, LOCKED BY DEFAULT (automatic Compliance-mode Vault Lock) — no separate Vault Lock configuration step needed
EncryptionEncrypted with an AWS-owned KMS key by default (protecting against accidental/malicious deletion of a CUSTOMER-managed key being used as an attack vector); as of November 2025, customer-managed KMS keys (CMKs) are also supported for organizations needing that level of control
Cross-account sharingShared to other accounts (including across an entire AWS Organization) via AWS Resource Access Manager (RAM), enabling DIRECT restore from the shared account without first copying the backup elsewhere
Service coverage growthEC2, EBS, RDS, S3, EFS at preview (2023); expanded to single-step cross-Region database snapshot copy for Aurora, Neptune, and DocumentDB (Feb 2026); expanded to support Amazon EKS (March 2026)
  • The single-step cross-Region database copy (Feb 2026) eliminated a previously required two-step process — copy to the target Region's standard vault FIRST, then copy again to the LAG vault. The old way required custom Lambda/scripts just to monitor intermediate copy status; the new single-action copy achieves faster RPOs and removes the intermediate cost entirely.
  • "Protect our backups so that even if our production account is fully compromised by ransomware with admin-level access, the backup data in a separate account remains intact and directly restorable" → Logically Air-Gapped Vault shared via RAM to an isolated recovery account, not just Vault Lock alone.
Multi-Party Approval (MPA) for LAG Vaults High — exam-fresh, June 2025
What it doesRequires multiple authorized individuals to approve critical recovery operations (like sharing/restoring from a LAG vault) BEFORE they execute — a distributed decision-making control preventing any single person from unilaterally authorizing recovery
The specific gap it closesAuthorizes access to backups for approved accounts EVEN WHEN the vault-OWNING account itself becomes inaccessible due to an inadvertent or malicious event — solving the "what if the account that owns the air-gapped vault is itself the one that got compromised/locked out" problem
MechanismApproval teams are managed through an IAM Identity Center-enabled Approval portal; team members review and approve sharing/recovery requests collaboratively
CostNo additional charge for integrating and using MPA teams with LAG vaults
  • This directly supports a "clean recovery account" pattern: provision a separate, pristine recovery account in advance, and use MPA-governed sharing requests to authorize restoring into it when needed — without depending on the (possibly compromised) original owning account's continued availability or trustworthiness.
AWS Elastic Disaster Recovery (DRS) Continuous replication + failover
What it doesContinuously replicates EC2-based workloads (and on-premises/other-cloud servers) at the block level to a recovery Region, enabling fast (minutes) recovery instance launch during an actual disaster
Cross-Region failbackAfter failing over to a recovery Region, DRS supports failing BACK to the original source Region — but failback requires starting REVERSED replication first (copying data from the recovered instances back to the source), which takes time and incurs cross-Region data transfer cost
Cross-account failbackAlso supported between AWS accounts — requires Failback and in-AWS right-sizing IAM roles to be pre-created via the Trusted Account page, in BOTH directions (source-to-recovery AND recovery-to-source)
Opt-in Region constraintIf either the source or recovery Region is an opt-in Region, that Region must be explicitly enabled in BOTH the source and recovery accounts; if both are opt-in Regions, both must be enabled in both accounts
  • Recovered instances after failover are NOT automatically protected going forward — to protect them, you must navigate to the recovery instance and (in a real DR event, as opposed to a drill) begin the reversed-replication/failback protection process deliberately.
  • Traffic redirection (e.g. via Route 53) to the newly failed-back instances is NOT handled by DRS itself — it's a separate step using your own DNS/traffic management tooling.
  • Cross-PARTITION failback (e.g. commercial ↔ GovClous) is explicitly NOT supported; cross-Region failback WITHIN the GovCloud partition (us-gov-west-1 ↔ us-gov-east-1) is supported.

AWS Exam Thinking

Requirement → Keywords → Expected Answer → why every distractor fails.

Protect backups so they remain intact and restorable even if production account credentials are fully compromised by ransomware
ransomware-resilientsurvives account compromisecross-account isolation
Expected Answer

Logically Air-Gapped Vault shared via RAM to a separate recovery account

DistractorWhy it's wrong
Vault Lock alone in the production accountProtects against deletion/retention-shortening within IAM rules, but the data still lives in the SAME, now-compromised account — doesn't achieve the cross-account isolation the requirement needs
Standard backup vault with cross-Region copy onlyStill within the same account — protects against a Regional outage, not an account-level compromise
Enforce a retention period that not even the account's own root user can shorten
retention enforcementnot even root
Expected Answer

Vault Lock in Compliance mode

DistractorWhy it's wrong
Vault Lock in Governance modeAuthorized IAM principals CAN still remove the lock or modify retention — doesn't meet the "not even root" bar
Authorize backup recovery access even though the original vault-owning account has become completely inaccessible
owning account inaccessibleauthorize recovery anyway
Expected Answer

Multi-Party Approval for the Logically Air-Gapped Vault

DistractorWhy it's wrong
Standard RAM sharing aloneRAM sharing alone doesn't provide a governed approval mechanism specifically designed to function when the OWNING account itself is inaccessible — MPA's approval team/portal is the purpose-built solution
Achieve a fast (minutes-scale) failover for EC2-based workloads to a different AWS Region during an actual disaster
fast failoverEC2 workloadsminutes-scale recovery
Expected Answer

AWS Elastic Disaster Recovery (DRS)

DistractorWhy it's wrong
AWS Backup with cross-Region copy aloneBackup/restore is generally slower than DRS's continuous block-level replication purpose-built for fast EC2 failover; Backup is point-in-time recovery, not continuous near-real-time replication
Reduce RPO and eliminate intermediate copy costs when sending Aurora database snapshots to a cross-Region air-gapped vault
single-step copyAuroracross-Region LAG vault
Expected Answer

Single-action database snapshot copy directly to the cross-Region LAG vault (Feb 2026 capability)

DistractorWhy it's wrong
Two-step copy (target Region vault, then LAG vault) with a custom Lambda monitorThe OLD, pre-Feb-2026 pattern — now unnecessary; the single-action copy achieves the same outcome faster and cheaper, without custom monitoring scripts

Integrations

AWS Resource Access Manager (RAM)
WhatThe mechanism for sharing a Logically Air-Gapped Vault to other accounts (including cross-Organization) for direct restore
Amazon GuardDuty
WhatMalware scanning of backups before restore completes, catching ransomware/malware captured into a recovery point
AWS KMS
WhatEncrypts LAG vault recovery points — AWS-owned key by default, customer-managed key (CMK) supported since November 2025, from the same account or across accounts
IAM Identity Center
WhatPowers the Multi-Party Approval portal where approval team members review and authorize LAG vault sharing/recovery requests
AWS Organizations
WhatBackup policies centrally enforce consistent backup plans across every account; LAG vaults can be shared across an entire Organization
AWS Backup Audit Manager
WhatProvides controls specifically for cross-Region copy, cross-account copy, and Vault Lock compliance — generating audit evidence of resilience posture

Costs, Limits & Quotas

Pricing Model

AWS BackupPer-GB stored (warm/cold tiers), plus restore charges
Cross-Region/cross-account copyStandard data transfer charges apply for the copy operation
Multi-Party ApprovalNo additional charge to integrate/use with LAG vaults
AWS DRSPer-hour charge per replicated source server, plus underlying storage/compute for replication and recovery instances; cross-Region failback transfer incurs standard data transfer cost

Common Cost Mistakes

Cost Optimization

Limits & Quotas

Vault Lock compliance grace periodTypically 72 hours — after which the lock is permanent for the life of any recovery points in the vault
LAG vault service coverage (preview-era baseline)EC2, EBS, RDS, S3, EFS — since expanded to include Aurora/Neptune/DocumentDB cross-Region single-step copy (Feb 2026) and EKS (March 2026)
DRS cross-partition failbackNOT supported between commercial and GovCloud partitions; cross-Region failback WITHIN GovCloud (us-gov-west-1 ↔ us-gov-east-1) is supported
DRS opt-in Region requirementAny opt-in Region involved in failover/failback must be explicitly enabled in BOTH the source and recovery accounts
⚠️ Exam trap

After a DRS failover, the newly recovered instances are NOT automatically protected against a further incident — reversed replication/failback protection must be deliberately initiated. This is a common "what's the very next step after failover" trap.

Best Practices & Common Exam Traps

8 — Best Practices

Must Know
  • LAG vaults are immutable by default — no separate Vault Lock step required
  • Vault Lock Compliance mode = immutable even from root after the grace period; Governance mode is removable by authorized IAM
  • LAG vault cross-account sharing (via RAM) is what actually achieves ransomware/compromised-account resilience — Vault Lock alone doesn't move data out of the production account
  • MPA solves recovery authorization even when the vault-owning account itself is inaccessible
  • DRS failover ≠ automatic ongoing protection — reversed replication must be deliberately started afterward
Good Practice
  • Use Organizations backup policies to centrally enforce consistent backup coverage across all accounts
  • Copy critical backups to a LAG vault in a dedicated, isolated recovery account
  • Enable GuardDuty malware scanning before restore for any backup that could have captured an active compromise
  • Conduct regular DRS failover AND failback drills, not just failover drills
Advanced Practice
  • Provision a "clean recovery account" in advance and govern access to it via Multi-Party Approval-backed LAG vault sharing
  • Use customer-managed KMS keys on LAG vaults (since Nov 2025) where regulatory requirements demand it, over the AWS-owned-key default
  • Use the single-action cross-Region database snapshot copy for Aurora/Neptune/DocumentDB to simplify and accelerate the cross-Region LAG vault pattern
  • Pre-create cross-account DRS failback IAM roles in BOTH directions well before an actual incident, not during one

9 — Common Exam Traps

MisconceptionReality
"Vault Lock alone makes our backups ransomware-proof"It prevents deletion/retention-shortening, but the data still lives in the same (potentially compromised) account — a Logically Air-Gapped Vault shared cross-account is what achieves true isolation
"Governance mode and Compliance mode provide the same level of immutability"Governance mode is removable by authorized IAM principals; Compliance mode becomes permanent (not even root can override) after the grace period
"LAG vaults require manually configuring Vault Lock after creation"False — LAG vaults come with automatic Compliance-mode lock by default, no separate configuration step
"After a DRS failover, our workload is automatically protected against a second incident"False — reversed replication/failback protection must be deliberately initiated; recovered instances are unprotected until then
"DRS failback works the same way across any two AWS partitions"False — cross-partition failback (commercial ↔ GovCloud) is explicitly unsupported; only cross-Region failback within the same partition (including within GovCloud) is supported

Lookalike Comparisons

ComparisonWhat actually differs
Standard vault vs Vault Lock vs LAG vaultReachable/deletable baseline vs WORM-protected (same account) vs immutable AND cross-account-isolated by default — increasing levels of resilience against the same-account compromise threat
Vault Lock Governance vs Compliance modeRemovable by authorized IAM principals vs permanently immutable (not even root) after the grace period expires
RAM sharing alone vs Multi-Party ApprovalGrants cross-account access to the resource generally vs specifically governs and authorizes RECOVERY requests, including when the owning account itself is inaccessible
AWS Backup vs AWS DRSPoint-in-time recovery points on a backup schedule vs continuous, near-real-time block-level replication purpose-built for fast EC2 failover — different RPO/RTO profiles for different workload criticality
DRS failover vs failbackLaunching recovery instances in the target Region/account during a disaster vs the structured, reversed-replication process to return to the original source afterward — failback is NOT just "failover in reverse," it has its own distinct prerequisites

Flashcards — 16 Cards

Click card to flip. Mark right or wrong to track score.

Click to reveal answer
1 / 16
Mark:   Score: 0/0

Practice Quiz — 9 Questions

SCS-C02 scenario style, Easy → Specialty. Select an answer to reveal the explanation.

out of 9 correct