Snowflake vs traditional data warehouses - CLI Comparison

Introduction

Traditional data warehouses store data on fixed hardware and can be slow to scale. Snowflake is a cloud data warehouse that separates storage and computing, making it faster and easier to scale without managing hardware.

When you want to quickly scale your data storage and processing without buying new servers.

When you need to run many users' queries at the same time without slowing down.

When you want to pay only for the computing power you use, not for idle servers.

When you want to store all your data in one place and access it easily from anywhere.

When you want automatic backups and security without manual setup.

Commands

Connect to Snowflake using the SnowSQL command line tool with your account, user, warehouse, database, and schema details.

Terminal

snowsql -a myaccount -u myuser -w mywarehouse -d mydatabase -s public

Expected Output

Welcome to Snowflake

You are now connected to Snowflake as user 'myuser'.

→

-a - Specifies the Snowflake account name

→

-u - Specifies the username

→

-w - Specifies the virtual warehouse to use

→

-d - Specifies the database to use

→

-s - Specifies the schema to use

Create a virtual warehouse in Snowflake that handles compute resources. It automatically suspends after 60 seconds of inactivity and resumes when needed.

Terminal

CREATE WAREHOUSE mywarehouse WITH WAREHOUSE_SIZE = 'SMALL' WAREHOUSE_TYPE = 'STANDARD' AUTO_SUSPEND = 60 AUTO_RESUME = TRUE;

Expected Output

Statement executed successfully.

Create a simple table in Snowflake to store customer data.

Terminal

CREATE TABLE customers (id INT, name STRING, email STRING);

Expected Output

Statement executed successfully.

Insert a row of data into the customers table.

Terminal

INSERT INTO customers VALUES (1, 'Alice', '[e-mail]');

Expected Output

1 row inserted.

Query the customers table to see the data you inserted.

Terminal

SELECT * FROM customers;

Expected Output

ID | NAME  | EMAIL

1  | Alice | [e-mail]

Key Concept

Snowflake separates storage and compute so you can scale and pay for each independently, unlike traditional data warehouses.

Common Mistakes

Trying to scale compute and storage together as one unit.

Traditional warehouses tie storage and compute, so scaling one forces scaling the other, wasting resources.

Use Snowflake's separate scaling to adjust compute power without changing storage size.

Not setting AUTO_SUSPEND on warehouses.

Without auto suspend, warehouses keep running and cost money even when idle.

Set AUTO_SUSPEND to a low value like 60 seconds to save costs.

Using a single warehouse for all workloads causing slow queries.

One warehouse can get overloaded, slowing down all queries.

Create multiple warehouses for different workloads to isolate and speed up queries.

Summary

Snowflake uses virtual warehouses to separate compute from storage for flexible scaling.

You create warehouses and tables with simple SQL commands in Snowflake.

Auto suspend and resume features help save costs by stopping compute when not in use.

Practice

1. What is a key advantage of Snowflake compared to traditional data warehouses?

easy

A. It is cloud-based and easy to scale on demand

B. It requires physical hardware setup

C. It has fixed resource limits that cannot be changed

D. It needs manual software installation on servers

Solution

1. Step 1: Understand Snowflake's deployment model

   Snowflake is built on the cloud, so it does not require physical hardware or manual installations.

2. Step 2: Compare with traditional warehouses

   Traditional warehouses often need physical setup and fixed resources, limiting scalability.

3. Final Answer:

   It is cloud-based and easy to scale on demand -> Option A

4. Quick Check:

   Cloud-based and scalable [OK]

Hint: Cloud means easy scaling and no physical setup [OK]

Common Mistakes:

• Thinking Snowflake needs physical hardware
• Assuming resources are fixed in Snowflake
• Confusing manual installation with cloud services

2. Which of the following is the correct way to describe Snowflake's resource usage?

easy

A. Snowflake uses only on-premises servers for compute

B. Snowflake requires upfront purchase of fixed compute resources

C. Snowflake does not support scaling compute resources

D. Snowflake charges based on actual usage, scaling compute as needed

Solution

1. Step 1: Review Snowflake's billing model

   Snowflake charges customers based on the compute and storage they actually use, allowing flexible scaling.

2. Step 2: Contrast with fixed resource models

   Traditional warehouses often require buying fixed compute capacity upfront, unlike Snowflake.

3. Final Answer:

   Snowflake charges based on actual usage, scaling compute as needed -> Option D

4. Quick Check:

   Pay for what you use [OK]

Hint: Snowflake bills by usage, not fixed resources [OK]

Common Mistakes:

• Thinking Snowflake requires upfront fixed resource purchase
• Believing Snowflake cannot scale compute
• Assuming Snowflake uses only on-premises servers

3. Given the following scenario: A company runs a traditional data warehouse with fixed compute resources. They experience a sudden spike in data queries. What is the likely outcome compared to using Snowflake?

medium

A. The traditional warehouse will automatically scale compute to handle the spike

B. Snowflake can scale compute instantly to handle the spike, traditional cannot

C. Both systems will fail to handle the spike due to fixed resources

D. Traditional warehouses handle spikes better because of fixed resources

Solution

1. Step 1: Understand traditional warehouse limitations

   Traditional warehouses have fixed compute capacity and cannot scale instantly to spikes.

2. Step 2: Understand Snowflake's scaling ability

   Snowflake can quickly add compute resources on demand to handle spikes in queries.

3. Final Answer:

   Snowflake can scale compute instantly to handle the spike, traditional cannot -> Option B

4. Quick Check:

   Instant scaling = Snowflake advantage [OK]

Hint: Only Snowflake scales instantly for spikes [OK]

Common Mistakes:

• Assuming traditional warehouses auto-scale
• Thinking fixed resources handle spikes better
• Believing both systems fail equally

4. A company tries to reduce costs by running their traditional data warehouse 24/7 at full capacity. What is a key problem with this approach compared to Snowflake?

medium

A. They pay for unused compute during low demand times

B. Snowflake requires running 24/7 at full capacity too

C. Traditional warehouses automatically pause when idle

D. Snowflake cannot pause compute resources

Solution

1. Step 1: Analyze traditional warehouse cost model

   Traditional warehouses have fixed compute running constantly, so costs remain high even when idle.

2. Step 2: Compare with Snowflake's cost efficiency

   Snowflake can pause compute when not in use, saving costs during low demand.

3. Final Answer:

   They pay for unused compute during low demand times -> Option A

4. Quick Check:

   Fixed compute costs even when idle [OK]

Hint: Traditional pays always; Snowflake pauses to save [OK]

Common Mistakes:

• Thinking Snowflake must run 24/7
• Believing traditional warehouses pause automatically
• Assuming Snowflake cannot pause compute

5. A company wants to migrate from a traditional data warehouse to Snowflake. Which of the following best describes a benefit they will gain in terms of management and cost?

hard

A. They will need to manage physical hardware but save on software licenses

B. They must buy fixed compute capacity upfront but get better performance

C. They reduce management effort and pay only for the compute and storage they use

D. They lose flexibility but gain better control over physical resources

Solution

1. Step 1: Understand Snowflake's cloud benefits

   Snowflake removes the need to manage physical hardware and automates many management tasks.

2. Step 2: Understand Snowflake's cost model

   Snowflake charges based on actual compute and storage usage, avoiding upfront fixed costs.

3. Final Answer:

   They reduce management effort and pay only for the compute and storage they use -> Option C

4. Quick Check:

   Less management + pay-as-you-go [OK]

Hint: Cloud means less management and pay for usage [OK]

Common Mistakes:

• Thinking physical hardware management is still needed
• Assuming fixed upfront compute purchase
• Believing flexibility is lost after migration