Independent DPDP educationBrowser-only workspace · no accounts, analytics or submissions

Cloud infrastructureSystem guide · implementation aid

Amazon Web Services: follow the evidence.

Connect services, regions, identities, logs, snapshots and restore paths to each data class.

00 What usually flows through Amazon Web Services

A working data journey, not a legal conclusion.

Application records, object storage, databases, logs, snapshots, backups, identities, support events and cross-account transfers. Replace this editorial model with the systems, data and roles your organisation can actually evidence.

Typical Amazon Web Services data journeySix stages show information moving from collection through operational systems, evidence and an end-of-lifecycle decision.010203040506
  1. 01ApplicationAPI · event · upload
  2. 02ServicesDB · object · queue
  3. 03Accessrole · key · account
  4. 04Copieslog · replica · pipeline
  5. 05EvidenceCloudTrail · config · alert
  6. 06Enddelete · snapshot · restore
Illustrative operating model. Verify actual fields, destinations, contracts, regions, retention and access.

01

Accountability

Role map

Roles depend on who decides purpose and means, who acts on instructions, and what each party actually does. Record the conclusion and its evidence; do not infer it from the vendor label.

  • Name workload, platform, security, account and data owners.
  • Separate organisation, account, role, service and support access.
  • Record AWS and other managed-service roles from current agreements and architecture.

02

Collection surfaces

Collection and notice touchpoints

Walk each entry path as a real user. Save the notice, fields, choices, time and destination rather than relying on a policy page alone.

  • Application APIs, databases, object storage, queues and logs.
  • Developer access, support cases, observability and data pipelines.
  • Backups, snapshots, replication, analytics and cross-account sharing.

03

Purpose discipline

Purpose and data minimisation checks

Every field and copy should have a named operating reason, accountable owner and review event.

  • Map data classes to only required services, fields, logs and regions.
  • Avoid personal data in resource names, tags and verbose logs.
  • Limit non-production copies and broad data-lake ingestion.

04

Choice evidence

Consent and preference evidence

When consent is relied on, preserve the affirmative action and withdrawal path. Where another legal route is assessed, record that analysis instead of manufacturing a consent record.

  • Carry collection and purpose evidence from the application layer.
  • Do not treat an IAM permission as a data principal’s consent.
  • Ensure downstream analytics uses match the recorded purpose.

05

Least privilege

Access control and privileged roles

Test ordinary view, sensitive fields, bulk action, export, configuration and integration access separately.

  • Review root use, IAM Identity Center, roles, policies, keys and cross-account trust.
  • Separate control-plane and data-plane access.
  • Exercise key rotation, session revocation and departure.

06

Lifecycle

Retention, deletion, backup and export behaviour

A delete button is not a lifecycle rule. Record the start event, end event, exception, system action, residual copy and accountable approver.

  • Define deletion for live stores, versions, replicas, caches, snapshots and backups.
  • Test restore paths so deleted data is not silently reintroduced.
  • Record region, archive and retention configuration by service.

07

Service chain

Processor, sub-processor and contract checks

Use the current contract and actual architecture. A product page cannot establish the complete role allocation for your organisation.

  • Retrieve current AWS terms, DPA, service terms and sub-processor information.
  • Inventory managed services, SaaS tools, support access and cross-account parties.
  • Record exit, deletion, incident and assistance expectations.
Open the processor-register guide

08

Detection + response

Logs, monitoring and breach evidence

Coverage, event types, retention and exportability vary. Preserve an evidence timeline without claiming that one log proves the complete event.

  • Confirm CloudTrail event coverage, trails or event stores, region scope and retention.
  • Include data events where the risk and service require them.
  • Protect logs and connect detections to the investigation workflow.

09

Request workflow

Rights-request search, export, correction and erasure workflow

  1. 01

    Translate a verified person identifier into application and storage keys.

  2. 02

    Search live services and owned data pipelines.

  3. 03

    Apply exceptions and execute application-controlled correction or deletion.

  4. 04

    Track versions, replicas, snapshots, backups and downstream confirmations.

10

Bounded configuration

Configuration checklist

Service-region map

Every relevant service and region has an owner and data class.

Admin path
Verify in the current admin console
Evidence to save
Architecture inventory and review date.
IAM access

Human, service and cross-account access are least-privileged.

Admin path
Verify in the current admin console
Evidence to save
Role/policy review and exceptions.
CloudTrail coverage

Required control- and data-plane events are retained deliberately.

Admin path
Verify in the current admin console
Evidence to save
Trail/event-store configuration and event sample.
Backup deletion

Versions, replicas and restore paths follow a documented lifecycle.

Admin path
Verify in the current admin console
Evidence to save
Backup map and restore/deletion exercise.

No menu-path fiction: open the current vendor documentation and your live console together. Feature names, paths and entitlements can change.

11

Retrievable proof

Evidence to save

Service, region and data-class map

IAM and cross-account review

CloudTrail coverage and sample

Encryption/key ownership record

Snapshot and backup lifecycle

Rights-search and restore exercise

Save redacted configuration evidence in an approved internal location. This private preview does not accept uploads or store these records.

12

Do not overclaim

Known limitations and questions for the vendor

Known limitations

  • CloudTrail event history is not the same as a configured long-term trail or event data store.
  • Service-specific logging and deletion behaviour varies.
  • Application identifiers and business logic remain the organisation’s responsibility.

Questions to resolve

  1. Which accounts, services and regions hold each data class?
  2. Are required data events enabled?
  3. Which snapshots and replicas can restore deleted records?
  4. Which external accounts and support paths can access data?

13

Traceable record

Official vendor sources, DPDP sources and corrections

Vendor documentation supports configuration questions only. DPDP statements are mapped separately to official Indian sources and phased commencement records.

Editorial status

Reviewed · not counsel-reviewed · educational implementation guidance, not legal advice, certification or a legal conclusion.

Report or inspect a correction

Amazon Web Services and AWS are trademarks of Amazon.com, Inc. or its affiliates. References are nominative and do not imply endorsement. No affiliation or endorsement is implied.

14 Continue the workspace

Turn this system review into owned gaps.

The browser-only readiness map can capture what is evidenced, partial, unknown or missing. It does not produce a pass, certificate or legal conclusion.

Answer the linked readiness itemTrace SEC-02 to evidence Read: Cross-border processing
Related system guides

Microsoft AzureCloud infrastructure

All system guidesSearch the complete stack