Study SnowPro DEA-C02 Warehouse Sizing: key concepts, common traps, and exam decision cues.
Compute questions are often really workload-shape questions. The exam wants the warehouse or serverless option that matches concurrency, latency, and operational burden.
| Requirement | Better first instinct |
|---|---|
| stable analytic workload with predictable demand | warehouse fit and sizing |
| concurrency pressure from many users or workloads | consider multi-cluster behavior |
| Snowflake-managed execution with less operational tuning | serverless feature fit |
| warehouse pain with no evidence yet | inspect workload pattern before resizing blindly |
| Workload clue | Stronger first answer |
|---|---|
| predictable analytic demand | warehouse sizing and fit |
| many concurrent users or tasks | concurrency or multi-cluster fit |
| team wants less compute administration | serverless fit |
The exam usually punishes answers that resize first and classify later.
| If the stem says… | Strong reading |
|---|---|
| “design scalable compute solutions” | match compute to workload shape |
| “concurrency” | multi-cluster or workload isolation may matter |
| “reduce operations burden” | Snowflake-managed or serverless options may fit better |
Serverless can be strong when the real requirement is lower operational burden. It is weak when the question is really about understanding the workload shape, concurrency pattern, or explicit compute ownership. The right answer still depends on the operational need, not on a general preference for managed services.
| Trap | Better rule |
|---|---|
| increasing size without understanding concurrency | size and concurrency are not the same issue |
| collapsing every workload onto one warehouse | isolation can reduce blast radius |
| ignoring Snowflake-managed compute options when the scenario rewards lower ops burden | fit the control surface to the team and workload |
| Scenario clue | Stronger answer shape |
|---|---|
| “predictable steady analytic workload” | warehouse fit and sizing |
| “queueing from many simultaneous consumers” | concurrency / multi-cluster reasoning |
| “team wants less operational tuning” | serverless option may fit |
| “compute pain but no evidence yet” | inspect workload shape before resizing blindly |
Compute-fit questions usually start by identifying whether the pressure is concurrency, workload shape, or operational overhead. If many users are queueing, think concurrency and multi-cluster fit rather than only warehouse size. If Snowflake-managed execution reduces operational burden and still fits the job, serverless may be stronger. The weak answer usually ignores concurrency and talks only about storage or generic scaling.