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: CloudWatch, CloudTrail, Config and incident investigation

AWS CloudTrail explained: activity history, data events and CloudWatch

CloudTrail records AWS account activity as events. Records help investigate the identity, time and requested operation. Collection depends on event type and configuration, and an operation name alone does not prove success or establish an incident cause.

What you will learn

Match an activity to its event category, distinguish Event history from trails and other stores, and interpret identity, operation, outcome and time.

Prerequisite

An AWS API requests a service operation. Distinguish resource management from access to data. Read fictional fields here; no account connection, log ingestion or evidence deletion is performed.

Distinguish account activity from service health

Use CloudTrail and CloudWatch according to the question
QuestionEvidence to examineBoundary
When and under which identity was a configuration operation requested?Relevant CloudTrail eventsCheck collection scope, outcome, account and Region
Did latency or error rates change?Metrics and application logs, including CloudWatchCheck subject, aggregation and collection
Did a change cause a specific failure?Correlate the change with requests, paths and before/after evidenceEarlier timing alone does not prove causation
Investigate the operation, outcome and related observations
Investigate the operation, outcome and related observations1: Operation request — When covered by the event category and collection — Recorded event. 2: Recorded event — Check outcome separately from its effects — Effect investigation12Operation requestRecorded eventEffectinvestigationInvestigate the operation, outcome and related observations1: Operation request — When covered by the event category and collection — Recorded event. 2: Recorded event — Check outcome separately from its effects — Effect investigation12OperationrequestRecorded eventEffectinvestigation
  1. Operation request → Recorded eventWhen covered by the event category and collection
  2. Recorded event → Effect investigationCheck outcome separately from its effects
Operation request
Identity and requested action
Recorded event
Read subject, time and outcome
Effect investigation
Correlate configuration and requests

Understand the limits of Event history first

Event history shows the past 90 days of management events for an account and AWS Region. Management activity includes resource creation, configuration and deletion. Relevant console and SDK activity can be recorded as well as CLI activity; consult the service-specific event documentation.

Management and data operations do not share one collection scope
ExampleCategoryInvestigation check
Request to change an S3 bucket policyManagement eventCheck the operation, Region and time range
S3 GetObjectData eventCheck whether a trail or other configured store collects the needed data events
Every application HTTP request or internal logNot a complete request log supplied by CloudTrail aloneCheck application/ingress logs and collection separately

Trails and existing customers’ CloudTrail Lake event data stores are separate from Event history. CloudTrail Lake closed to new customers on May 31, 2026; CloudTrail and trails remain available. Do not plan on new Lake enrollment. Review the existing configuration and current alternatives such as CloudWatch, alongside collection, destination, retention, access and costs. Preserve existing investigation evidence when considering configuration changes. This lesson creates or changes none of these resources.

T1: read the operation together with its failed outcome

T1: a denied request to change a bucket policy
FieldFictional value
eventTime2026-10-24T11:58:00Z
eventSource / eventNames3.amazonaws.com / PutBucketPolicy
awsRegionap-northeast-1
userIdentity.typeAssumedRole
userIdentity.arnarn:aws:sts::111122223333:assumed-role/ExampleChangeRole/demo-session
errorCodeAccessDenied

PutBucketPolicy identifies the requested operation in T1. AccessDenied records a denied request in this example. Its presence does not mean the policy was changed. Determine actual effects from the subject, outcome, related records and relevant configuration.

AssumedRole identifies activity using role-session credentials. A session name at the end of an ARN does not by itself authenticate a particular human. Correlate it with trusted role-assumption and authorized identity/session records.

Investigate within a defined evidence scope

  1. Classify the activity as management, data or an application request.
  2. Fix account, Region, time range, resource and read authority.
  3. Verify collection and retention for the needed event category.
  4. Read identity, action, subject, time and outcome separately from effects.
  5. Correlate configuration/request evidence and retain unresolved collection gaps.

In actual investigations, restrict access to the required subjects and time range and handle identity/source information appropriately. Lack of read access does not justify broad administrator grants or sending secrets into a learning exercise.

Practice reading scope and outcome

Choose an option to see the result and explanation.

S3 GetObject is absent from Event history. What follows?
Read the answer and reason

Answer: Check data-event collection and its destination · Verify that the required event category is collected before interpreting the records.

What do PutBucketPolicy and AccessDenied establish in T1?
Read the answer and reason

Answer: The change request was denied · Read the operation with its outcome; this example does not establish a successful change.

What do AssumedRole and the session name establish alone?
Read the answer and reason

Answer: Investigate it as role-session activity · Correlate trusted identity and assumption records within the required scope.

Which evidence primarily shows CPU or response-latency trends?
Read the answer and reason

Answer: Aligned metrics and relevant application logs · Examine measurements and logs with aligned subject, period and statistic.

Check, then explain in your own words

  1. Explain Event history’s event category and time window.
  2. Separate T1’s identity, requested operation and outcome.
  3. Explain when data events or application logs are needed.
Check your understanding: What should you distinguish from the presence of an operation event?

Collection scope, success/failure, actual effects and incident causation.

An action name or earlier timestamp alone does not prove a successful change or the cause of a particular failure.

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.