Study DEA-C01 Programming, IaC and Code Performance: key concepts, common traps, and exam decision cues.
DEA-C01 does not test language-specific syntax, but it does expect data engineers to reason about SDK usage, Lambda behavior, IaC, CI/CD, and simple code-performance choices inside pipelines.
Infrastructure as code: Defining infrastructure in versioned templates or code so environments can be created and updated repeatably.
Cold start: Initialization delay when a compute runtime such as Lambda starts a new execution environment.
Packaging strategy: How code, dependencies, and deployment configuration are assembled and shipped into a runtime environment.
DEA-C01 is usually not asking whether you can write the best code. It is asking whether you can keep a data platform maintainable and fast enough without inventing unnecessary custom logic.
| Requirement | Strongest first fit | Why |
|---|---|---|
| repeatable infrastructure deployment | infrastructure as code | DEA-C01 expects repeatable environment definition rather than console-only work |
| lightweight event-driven transformation or glue logic | AWS Lambda | The need is short-lived code execution tied to events or orchestration |
| direct interaction with AWS services from code | AWS SDK usage | The problem is service integration from code, not infrastructure templating |
| deployment discipline across environments | CI/CD plus IaC | The issue is controlled rollout and repeatability |
| slow runtime behavior due to wasteful scans, poor batching, or bad dependency packaging | code and runtime performance review | DEA-C01 expects practical efficiency reasoning, not only deployment vocabulary |
| If the stem emphasizes… | Think first | Why this fits |
|---|---|---|
| repeatable environments across dev, test, and prod | IaC | The control problem is environment drift, not just application code |
| direct service calls, enrichment logic, or API-driven pipeline behavior | SDK or code path | The logic belongs in code, not only in infrastructure templates |
| lightweight event-driven compute | Lambda | The execution pattern is short-lived and reactive |
| deployment reliability and rollback control | CI/CD plus IaC | The issue is safe delivery, not just function logic |
| memory pressure, timeouts, cold starts, or slow scans | runtime and code-performance tuning | DEA-C01 wants practical operational tuning, not just service naming |
flowchart LR
A["Pipeline change needed"] --> B{"What is the actual issue?"}
B -->|Environment drift| C["IaC"]
B -->|Custom service interaction or logic| D["Code / SDK"]
B -->|Short-lived event compute| E["Lambda"]
B -->|Safe rollout across stages| F["CI/CD + IaC"]
B -->|Slow runtime behavior| G["Performance tuning"]
| Situation | Better reading |
|---|---|
| a simple mapping or managed ETL feature already exists | use the managed transform unless the stem requires custom behavior |
| the logic needs custom branching, external API calls, or special formatting | code may be justified |
| the environment must be rebuilt consistently in dev, test, and prod | IaC is the stronger answer |
| runtime timeouts appear after dependency or batch-size growth | investigate packaging, memory, batch sizing, and runtime design |
| Symptom | Strong first review |
|---|---|
| Lambda cold starts increased | dependency size, initialization work, package strategy |
| records are processed too slowly | batching, parallelism, repeated calls, unnecessary scans |
| deployments differ across environments | template drift, manual changes, weak CI/CD discipline |
| a simple mapping now needs custom branching or external calls | code may be justified instead of forcing only managed transforms |
| Trap | Better reading |
|---|---|
| “IaC only matters to platform teams.” | DEA-C01 expects data platforms to be repeatable too. |
| “Lambda performance is only about language choice.” | Packaging, memory, batch sizing, and initialization overhead matter as well. |
| “If code works once in the console, deployment is solved.” | Repeatability and CI/CD discipline still matter. |
| “Custom code is always stronger than managed transforms.” | The exam often rewards the simpler managed option unless the requirement truly needs custom code. |
When the stem mixes code and operations, use this order:
A team has a small event-driven enrichment step, but deployments across environments keep drifting because resources are recreated manually in the console. Cold starts also increased after a dependency-heavy package was added. What is the strongest reading first?
Correct answer: B. DEA-C01 expects you to combine repeatable infrastructure discipline with practical runtime-performance reasoning.