What Is Snowflake?

Playback speed

Swipe, scroll or use ← →

A cloud-native data platform

Everything else in this certification (warehouses, roles, loading, sharing) sits on top of one idea: what kind of thing Snowflake actually is. Fix that first, and the rest of the vocabulary lands in the right place.

Snowflake is an advanced data platform delivered as a self-managed service. It brings together data storage, processing, and analytics and uses a SQL query engine with an architecture designed natively for the cloud. This is the first useful mental model: Snowflake is not a local database package or a single reporting screen. It is a service in which data and multiple kinds of work can meet.

A common mistake	The accurate picture
A database engine you install on a server	A data platform delivered to you as a service
A dashboard-only reporting product	Storage, processing, and analytics combined in one platform
A driver or client that holds your data	The service that drivers and clients connect to

1 / 19

Quick check

Which description best captures Snowflake?
1. AA self-managed cloud data platform that combines storage, processing, and analytics
  
  Right. Snowflake is the platform and the service, with storage, processing, and analytics integrated rather than assembled from unrelated local tools.
2. BA local database package dedicated to storing structured tables
  
  Snowflake is delivered as a cloud service and cannot be installed as a local database package.
3. CA client driver that supplies storage but leaves query processing elsewhere
  
  A driver is a way for an application to reach Snowflake; it is not the platform, and it neither stores data nor runs queries.
2 / 19
Where the platform runs

The location of the platform is fixed; you cannot choose to install it elsewhere.

Snowflake uses public-cloud infrastructure for virtual compute instances and persistent data storage. It cannot be installed and run locally or on private-cloud infrastructure, whether on-premises or hosted. You connect to the service instead of installing the platform on a private server.

That has a hard consequence in scenario questions. If a requirement says the whole platform must run on our own hardware with no public cloud, then no client changes the outcome. The web interface is a window onto the service, and a JDBC or ODBC driver is only a connection: neither one hosts persistent storage or compute.

3 / 19
Quick check

A regulated team requires the complete data platform to run on its own on-premises hardware and forbids public-cloud infrastructure. What conclusion follows?
1. AAn ODBC driver can host Snowflake storage and compute inside the private network
  
  A driver connects an application to the service; it cannot hold persistent data or process queries on its own.
2. BThat deployment requirement is incompatible with how Snowflake is delivered
  
  Right. Snowflake runs its virtual compute instances and persistent storage on public-cloud infrastructure and cannot be installed locally or on private-cloud infrastructure, hosted or on-premises.
3. CThe web interface can be installed locally so the platform runs on private hardware
  
  The web interface is one of several ways in to the service, not a copy of the platform that can be deployed privately.
4 / 19
What the service takes off your hands

"Self-managed service" is not marketing language: it names a concrete division of labour, and the exam tests which side of the line a task falls on.

Customers do not select, install, configure, or manage physical or virtual hardware for Snowflake. There is virtually no software for customers to install, configure, or manage. Snowflake handles ongoing maintenance, management, upgrades, tuning, software updates, and infrastructure.

Read that as a list of work that simply never reaches your team: no servers to provision, no patch weekends, no version rollout to plan, no hardware replacement when a node fails. For warehouse types that take a size, such as standard warehouses, your team still chooses the size that matches its workload.

5 / 19
Quick check

Which responsibility does Snowflake handle as part of its self-managed service?
1. ADeciding which employees should receive each business role in the organization
  
  Snowflake supplies access-control capabilities, but the customer decides which people hold which roles.
2. BWriting every analytical query that the business needs
  
  Snowflake processes queries; customers and their applications decide which queries to submit.
3. COngoing platform maintenance, upgrades, tuning, and software updates
  
  Right. Ongoing maintenance, management, upgrades, tuning, software updates, and infrastructure are handled by the service.
6 / 19
Keep your progress in the app

That’s 3 of 9 quick checks. In the app they stay answered, and every lesson remembers where you left off.

Opens in your browser. Nothing to install.
Start studying

On your phone, with the same account

Keep reading here

What stays your decision

The service boundary cuts both ways. A team still decides what data to use, how to model it, who may access it, and which workloads to run. The platform removes much of the underlying hardware and software administration so the team can concentrate on those data decisions.

Snowflake's side of the line	Your side of the line
Selecting and managing the hardware behind the service	Deciding which data the business needs and what it means
Installing and updating the platform software	Modelling that data and defining its quality rules
Ongoing maintenance, upgrades, and tuning	Deciding who may access which objects
Providing the compute that runs queries	Deciding which workloads and queries to run

The useful test is simple: does this task depend on knowing the business? If it does, it stays with you. If it is about keeping the platform running, it belongs to Snowflake.

7 / 19

Quick check

Which task remains a customer decision even though Snowflake is self-managed?
1. AChoosing what data to use and who should be allowed to access it
  
  Right. Self-managed describes the platform operations; the business meaning of the data and its authorization remain customer decisions.
2. BReplacing failed platform hardware used by Snowflake virtual compute
  
  Customers do not select or manage the physical or virtual hardware behind the service.
3. CPerforming the platform's ongoing infrastructure upgrades and tuning
  
  Ongoing maintenance, upgrades, and tuning are handled by Snowflake, not the customer.
8 / 19

Three forms of data

Snowflake classifies the data it supports by schema, not by file format or by the workload that will use it. Three forms cover everything.

Snowflake supports structured data with a strict tabular schema, semi-structured data with a flexible schema, and unstructured data with no inherent schema.

Form	Schema	Typical example
Structured	Strict and tabular	Rows and columns in a table
Semi-structured	Flexible	A JSON file or an XML file
Unstructured	None inherent	A document, an image, an audio file

Watch the trap in exam options: compressed, transactional, temporary, and public-cloud describe storage behaviour, object lifetime, or infrastructure, and none of them names a data form.

9 / 19

Quick check

Which set lists the three broad data forms supported by Snowflake?
1. AStructured, semi-structured, and unstructured data
  
  Right. The classification is by schema: strict tabular, flexible, or no inherent schema.
2. BRelational, temporary, and privately hosted data sets
  
  Temporary describes how long an object lives and private hosting describes deployment; neither names a form of data.
3. CTabular, public-cloud, and machine-learning data
  
  Public cloud describes the infrastructure and machine learning describes a workload, so only one of the three items is a data form.
10 / 19
Four integrated workload areas

Rather than treating each workload as a different product, begin by asking what job is being done with the data. Its integrated features support data engineering, analytics, AI and machine learning, and applications and collaboration.

Data engineering features help teams ingest, transform, and deliver data. Ingestion options include the COPY INTO command for files, Snowpipe for files as soon as they land, and Snowpipe Streaming for continuous row-level data; transformation options include dynamic tables, streams and tasks, and Snowpark.

Analytics features help extract insights, patterns, and trends for uses such as business intelligence and predictive modeling. Aggregate and window functions, common table expressions, and semantic views all serve this area.

11 / 19

Four integrated workload areas

Snowflake also provides AI and machine-learning capabilities that can work with current Snowflake data in a secure environment. Snowflake Cortex uses large language models to work with unstructured data, and Snowflake ML supports building your own models, both without moving the data to a separate platform.

Application and collaboration capabilities let teams build applications and share controlled access to data. Streamlit, Snowpark Container Services, and the Snowflake Native App Framework build applications; Secure Data Sharing, listings, and data clean rooms share access under control.

Area	The question it answers
Data engineering	How does the data get in, get reshaped, and get delivered?
Analytics	What do the insights, patterns, and trends say?
AI and ML	What can a model predict or understand from this data?
Applications and collaboration	What do we build on it, and who else may use it?

12 / 19

Quick check

Which workload area focuses on ingesting, transforming, and delivering data?
1. AAI and machine learning
  
  AI and machine learning use prepared data for model work; they do not name the ingest-transform-deliver responsibility.
2. BData engineering
  
  Right. Data engineering is the area whose features help teams ingest, transform, and deliver data.
3. CApplications and collaboration
  
  Applications and collaboration is about building applications and sharing controlled access, not about the ingestion pipeline itself.
13 / 19
Bringing the workloads to the data

Notice what the four areas have in common: they are features of one platform, not four products that each need their own copy of the data. That is the point of the design. Instead of moving data to a separate system for every team, you bring the workloads directly to the data.

Consider three teams: one ingests raw files, one analyses trends, and one needs to give a partner controlled access to a subset. With separate systems, teams may maintain synchronized databases, and these copies can drift apart. On one platform, the engineering, analytics, and collaboration features work against the same data foundation.

Instead of synchronizing separate databases for each team, you bring the different workloads directly to the data.

The same logic rules out the tempting shortcuts: a client driver is connectivity, not a repository, and limiting the platform to structured tables and dashboards would drop most of what the four areas cover.

14 / 19
Quick check

Three teams need to ingest data, analyze trends, and share controlled access while avoiding synchronized copies in separate systems. Which idea addresses the combined need?
1. AMove each workload onto its own separate local database installation
  
  Separate installations recreate exactly the fragmentation and copy-synchronization problem the team wants to avoid.
2. BUse a client driver as the shared persistent repository
  
  A driver provides connectivity between an application and the service; it is not a place where data persists.
3. CBring the integrated workloads to the data within Snowflake
  
  Right. The integrated feature areas let engineering, analytics, and collaboration work run against the same platform and the same data.
15 / 19

Ways to connect

A platform is only useful if people and programs can reach it. Snowflake connection choices include Snowsight, command-line clients, native APIs, drivers, native connectors, and third-party technologies.

Way in	Who uses it	Example
Snowsight	A person working in a browser	The Snowflake web interface
Command-line clients	Scripted or terminal work	Snowflake CLI
Native APIs	Code that manages resources	Python and REST APIs
Drivers	Applications that query Snowflake	JDBC, ODBC
Native connectors	Applications built around a platform	Apache Kafka, Apache Spark
Third-party technologies	Existing tooling	ETL and BI tools

These are ways into the same service, not separate Snowflake platforms. A web session and a driver session differ in client experience; the storage, the compute, and the data they reach are the platform's, not the client's.

16 / 19

Quick check

What is the relationship between Snowsight and a Snowflake driver?
1. AA driver is the Snowflake platform, while Snowsight is an unrelated reporting product
  
  Neither client surface is the platform, and Snowsight is one of the listed Snowflake connection choices.
2. BBoth connect to the same service, but one is a web interface and the other supports applications
  
  Right. Both are connection choices onto the same service; they differ in who uses them and how, not in the platform they reach.
3. CSnowsight holds persistent storage while the driver supplies the compute layer
  
  Storage and compute are capabilities of the platform, not responsibilities split between a web interface and a driver.
17 / 19
Key takeaways
- Snowflake combines storage, processing, and analytics as a self-managed, cloud-native data platform, reached over the public cloud and never installed locally or on private-cloud infrastructure.
- Snowflake manages the underlying hardware, software maintenance, updates, and infrastructure, while the business meaning of the data, its modelling, its access decisions, and its workloads stay with you.
- The platform supports structured, semi-structured, and unstructured data across engineering, analytics, AI, applications, and collaboration workloads: one platform the workloads come to, rather than one copy of the data per team.
- Snowsight, command-line clients, native APIs, drivers, native connectors, and third-party technologies are all doors into that same service.
18 / 19
Quick check

A team wants engineering, analytics, and AI workloads to use current data without maintaining separate platform software. Which Snowflake characteristic fits both constraints at once?
1. AA local SQL engine installed separately for each workload team
  
  Separate local installations conflict with the integrated-platform goal and with a service that cannot be installed locally at all.
2. BA dashboard service that leaves out ingestion and machine-learning work
  
  A dashboard-only service would omit the required data engineering and AI work, so it satisfies neither constraint fully.
3. CIntegrated workload features delivered through a self-managed data platform
  
  Right. The workload areas are integrated features of one platform, and the service handles the underlying software and infrastructure so no team maintains it.
19 / 19
9 quick checks · then the test

In the app, finishing the quick checks opens this lesson’s 10-question test, and the ones you miss come back exactly when you’re about to forget them.

Take this lesson’s test → Next lesson: Snowflake Structure and Core Vocabulary

Opens in your browser. Nothing to install.
Start studying

On your phone, with the same account