CV0-004 deep review — September 29, 2026

Same-context AI review; independent human review pending. Study guide.

Scope and lifecycle

The current Cloud+ page retains V4/CV0-004, September 24, 2024 launch, weights 23/19/17/19/10/12, maximum 90 questions, 90 minutes, score 750/900 and English/Japanese. Retirement remains estimated 2027 without an exact date or replacement announcement. Both monitored objective and lifecycle hashes are unchanged; no accepted snapshot was rewritten.

Read all numbered objectives and detailed bullets in the public document 5.0 PDF: 17 pages, 33 objectives in groups of 11/5/4/6/4/3, compared with 31 main-page summary rows. Page 9 layout confirms 3.1 observability, 3.2 scaling, 3.3 backup and 3.4 lifecycle. The receipt maps original paraphrases with published IDs; no PDF is redistributed.

Both official sources recommend 2–3 years as a systems administrator or cloud engineer. Removed the guide and catalog’s unsupported extra 12 months of cloud experience; the PDF’s Network+/Server+ or equivalent knowledge remains a recommendation. Added explicit 1.11 AI/ML and IoT, 2.1 community cloud, integration/tool roles and original design/evaluation questions.

Teaching and primary references

NIST SP 800-145 supplies foundational community/hybrid taxonomy, not a complete modern provider catalog. The PDF’s CLI expansion is qualified using the AWS command-line interface introduction. Provider-specific examples remain labeled; they do not redefine neutral exam scope.

Clarified Terraform sensitive-data display versus state/plan persistence and version/provider constraints. Validation checks structure; a plan checks a particular input/state context. Detailed plan codes distinguish unchanged, error and changes-present outcomes. The example deliberately uses Terraform 1.12.2, with only built-in input/triggers_replace behavior; newer online documentation features are not silently adopted.

AWS Budgets documents delayed information/notifications, so alerts are not an instantaneous spending cap. Invented cost, capacity, availability, IOPS-throughput and RTO/RPO cases state assumptions. HTTP status explanations use targeted RFC 9110 definitions and preserve the distinction between evidence and root cause, including refusal/non-disclosure and gateway failures.

Malcolm Featonby’s Builders’ Library article supplied request-identity, conflicting-intent and retention-contract reasoning. Its publication date was not confirmed. The guide applies those ideas to an original local model without claiming AWS internals, service retention or exactly-once delivery.

Executed evidence and limits

Two original public examples and a harness passed 61 checks using Terraform 1.12.2, Python 3.13.14 and SQLite 3.50.4. Terraform initialized only its built-in provider, validated a typed configuration, applied reviewed local plans, demonstrated no-change/update/replacement outcomes, rejected six invalid inputs and destroyed the local resource. No actual replica, image or cloud infrastructure was provisioned.

The SQLite receipt-and-effect example uses parameter binding and explicit SQL transactions. Checks covered repeated requests, different callers, conflicting payloads, prior-result semantics, exception rollback, reconnect, an owned worker exiting before commit and eight concurrent calls producing one new effect. Invalid amounts and empty caller input fail. The standalone in-memory example also ran; owned temporary database/state fixtures were cleaned.

No external Terraform provider, cloud API/credentials, queue, remote message, image pull, billed resource, workload rollout or complete cloud lab was used. A local database commit does not include external effects or queue acknowledgement. The receipt table has no expiry policy, and tests do not simulate arbitrary disk/power failures or establish distributed exactly-once behavior. All eight full labs remain proposed with stronger acceptance/negative-case evidence. There are 46 answered checks and three integrated scenarios.

Catalog and validation

Current metadata: Pluralsight six courses/14 hours, July–October 2024 dates and no verified lab count; LinkedIn Total Seminars 6h13/12 quizzes, December 1, 2025; MeasureUp product-specific 186 questions/March 2025, distinct from its generic approximate-150 FAQ. CompTIA lists Perform 30–60h, Learn 25–40h, Practice 10–20h and Labs 15–25h; these overlap. Other study-hour ranges are planning suggestions.

O’Reilly and Udemy blocked automated rechecking. Earlier fourth-edition/480-page/12h34 and 10h54/instructor/revision/assessment claims are qualified. Paid interiors, questions and labs were not accessed. Places to learn is final. Repository validation, 176 unit tests, strict site build, generated-site validation, catalog consistency and diff checks are recorded in the operational receipt after execution.