COF-C03 - Snowflake AI Data Cloud Features and Architecture (31% of the exam) - Section 1.1

Describe the three-layer Snowflake architecture of database storage, query processing, and cloud services, and how the layers scale independently.

Explain the three layers of the Snowflake architecture: centralised database storage, the multi-cluster compute (query processing) layer made of virtual warehouses, and the cloud services layer that coordinates authentication, metadata, optimisation, and security. Recognise that storage and compute scale and are billed independently, and how this differs from shared-disk and shared-nothing designs.

three-layer architecturedatabase storage layerquery processing layercloud services layerseparation of storage and compute

Practice question for this objective

Free sampleSnowflake AI Data Cloud Features and Architectureeasy

A data engineer is describing how the Snowflake architecture is structured. Which statement correctly identifies the three independent layers that make up the platform?

ADatabase storage, query processing using virtual warehouses, and cloud services that coordinate the whole system. Correct
BDatabase storage, an in-memory cache layer, and a reporting layer that renders dashboards for end users.
CA staging layer, a transformation layer, and a presentation layer arranged in a traditional ETL pipeline.
DA metadata layer, a security layer, and a billing layer that together meter and charge for usage.

Identify the three Snowflake layers as database storage, query processing, and cloud services. Snowflake separates storage, compute, and coordination into three distinct layers so each can scale on its own without affecting the others, which is the defining feature of the architecture.

Why A is correct: This is the documented three-layer architecture: storage, query processing through virtual warehouses, and cloud services, each scaling independently.

Why B is wrong: An in-memory cache and a reporting layer are not the named architectural layers, so this misstates the design even though caching does exist within compute.

Why C is wrong: These describe a generic ETL data-flow pattern, not the Snowflake platform layers, so the option is tempting but wrong for this question.

Why D is wrong: Metadata, security, and billing are functions handled inside cloud services, not three separate top-level layers, so this option is incorrect.

See more COF-C03 practice questions, answers explained.

Exam traps in Snowflake AI Data Cloud Features and Architecture

Answers that look right on this material and are not. Each one is a distractor from a different question in the COF-C03 bank for this domain.

The query processing layer, because the virtual warehouse handles parsing, planning, and authorisation as the first stage of running every query.
Why it is wrong: Tempting because warehouses run the query, but parsing, optimisation, and authorisation are coordinated by cloud services before compute scans data.
The query processing layer, because each virtual warehouse stores its own query results on local disk for later reuse by any session.
Why it is wrong: Tempting because warehouses keep a local-disk cache, but that cache holds scanned data and is lost on suspend, not the account-wide reusable result set.
Persisting table data as compressed, immutable micro-partitions within the centralised storage layer.
Why it is wrong: Persisting micro-partitions is the job of the database storage layer, so this describes storage rather than cloud services.

Examworthy is not affiliated with or endorsed by Snowflake. Original, blueprint-aligned practice material only.