These services tell you what is happening in and around your workloads, beyond the metrics and logs in CloudWatch. X-Ray traces single requests across services. AWS Health reports AWS’s own disruptions and planned changes that affect your accounts. Trusted Advisor checks your accounts against best practices. OpsWorks, a retired configuration management service, appears here only as a map for migrating off it.

Choosing a service

QuestionServiceSource of truthCost and access
Where is this request slow or failing?X-RayTraces your code sendsSampling rules control cost; traces kept 30 days
Is AWS itself having a problem, or changing something, that affects me?AWS HealthAWS’s events for your accountDashboard and EventBridge events free; the API needs a qualifying support plan
What in my account breaks best practice?Trusted AdvisorAutomated checks on your resourcesFull checks and the API need Business Support+ or above
How do I get off OpsWorks Stacks?OpsWorksYour existing stacksDiscontinued May 26, 2024

Health and Trusted Advisor both send changes to EventBridge, so route both into the same alerting path (SNS, Slack, or incident tooling) rather than checking consoles.12

AWS X-Ray

X-Ray records the path of each request through your front end, microservices, databases, and downstream AWS APIs. Each service contributes a segment, and each call within it a subsegment; a trace joins them into one request path, and the service map draws the calls between services with latency and errors.

flowchart LR
    accTitle: How trace data reaches X-Ray
    accDescr: The X-Ray SDK in each service records segments and subsegments for incoming requests, AWS SDK calls, HTTP calls, and database queries. It sends segment documents over UDP to the X-Ray daemon, which batches and uploads them to X-Ray, where traces and the service map are built.
    App[Service code with X-Ray SDK] -- segment documents over UDP --> D[X-Ray daemon]
    D -- batched upload --> X[X-Ray: traces and service map]
  • The SDKs (Java, Python, Node.js, Go, .NET, Ruby) create subsegments automatically for HTTP clients, AWS SDK calls, and database queries; instrument at service boundaries.
  • Run the daemon yourself on EC2 or on premises; Lambda and Elastic Beanstalk include it.
  • Sampling rules set how many requests are traced; lower them on high-traffic, low-value endpoints to control cost.3
SymptomCheck
No tracesSDK instrumentation, daemon status, and IAM permissions
Downstream calls missingSDK version, and that HTTP and AWS SDK clients are instrumented
Trace breaks between servicesTrace header propagation through proxies and Lambda

AWS Health

AWS Health reports service disruptions, scheduled changes, and account notifications (such as security and billing) that affect your resources, each with affected entities and guidance. The Health Dashboard needs no setup and shows only the current account. Every customer can receive the same events through EventBridge (source aws.health); the Health API, for integrating with other tools, needs Business Support+ or above.

  • Subscribe to Health events through EventBridge for all accounts and Regions.
  • Watch scheduled changes early, so maintenance windows are not a surprise.1

AWS Trusted Advisor

Trusted Advisor runs best-practice checks in five categories: cost optimization, security, fault tolerance, performance, and service limits. Examples include underutilized EC2 instances, MFA on the root account, RDS backups, and usage near a quota. The console shows each check as green, yellow, or red with a recommended action.

Support planChecksAPI and EventBridge
BasicService limits plus selected security and fault-tolerance checks; security checks refresh manuallyNo
Business Support+, Enterprise Support, or AWS Unified OperationsAll checksYes
  • Review on a schedule with an owner per recommendation, and act first on security and service-limit findings such as root MFA and open security groups.2

AWS OpsWorks

OpsWorks Stacks used Chef to configure EC2 instances: a stack per Region held layers of identically configured instances, and Chef recipes ran at lifecycle events (setup, configure, deploy, undeploy, shutdown), with auto-healing and load- or time-based scaling. It was discontinued for all customers on May 26, 2024.

OpsWorks pieceMigrate to
Instance operations and Chef recipesSystems Manager documents, or user data
Stacks and layers as infrastructureCloudFormation
Application deploymentsCodeDeploy
Whole application platformContainers or Elastic Beanstalk

Test the migration on a subset of workloads before decommissioning the old stacks.4

Footnotes

  1. AWS Health - Runbook & Reference, original ↩ ↩2

  2. AWS Trusted Advisor - Runbook & Reference, original ↩ ↩2

  3. AWS X-Ray - Runbook & Reference, original ↩

  4. AWS OpsWorks - Runbook & Reference, original ↩