---
title: 'What is Snowflake? Learn Snowflake in 30 Minutes'
source: 'https://youtube.com/watch?v=h78-o40MnRM'
video_id: 'h78-o40MnRM'
date: 2026-07-31
duration_sec: 1876
---

# What is Snowflake? Learn Snowflake in 30 Minutes

> Source: [What is Snowflake? Learn Snowflake in 30 Minutes](https://youtube.com/watch?v=h78-o40MnRM)

## Summary

This video provides a beginner-friendly introduction to Snowflake, a cloud-native data warehouse platform. It covers core concepts like data warehouses vs. relational databases, walks through a hands-on tutorial of loading and querying sample data, and explains key features and benefits. The tutorial also compares Snowflake with traditional data warehouses to highlight its advantages in modern data engineering.

### Key Points

- **Snowflake Overview and Popularity** [00:01] — Snowflake IPOed in 2020 raising $3.36 billion, the biggest software IPO. 751 of Forbes Global 2000 companies use Snowflake, and job portals show tens of thousands of Snowflake jobs.
- **Data Warehouse Architecture** [00:45] — Raw data from relational databases and unstructured sources is extracted into a data lake, then transformed (cleaned, aggregated) and loaded into a data warehouse. Snowflake is that analytics-ready data warehouse, supporting BI dashboards, Jupyter notebooks, and AI programs.
- **RDBMS vs Data Warehouse** [02:50] — RDBMS (Oracle, MySQL) is built for OLTP — many small, high-concurrency transactions. Data warehouses like Snowflake handle OLAP — few very large queries scanning billions of rows.
- **Snowflake History and Evolution** [04:08] — Snowflake started in 2012, became generally available as a cloud-native data warehouse, IPOed in 2020, and launched Snowpark to support Python and Java. Today it positions itself as an AI data cloud.
- **Setting Up a Free Snowflake Account** [05:47] — Go to snowflake.com, click 'Start for Free', register with an email, select a region, and activate via email. The free trial provides $400 credits and 30 days. Then click 'Start' to explore sample data.
- **SQL Worksheet Interface** [07:50] — Snowflake's worksheet is like a Jupyter notebook for SQL. You can run a single query (Ctrl+Enter) or all queries. Before working, set the context: role, warehouse, database, and schema.
- **Loading Sample Data from S3** [11:32] — Create a schema, create a table (e.g., menu), create an external stage pointing to an S3 bucket, list files in the stage, then use COPY INTO to load the data. After loading, SELECT COUNT(*) shows 100 rows.
- **Querying the Loaded Data** [16:28] — Run queries like SELECT TOP 10 * FROM menu, filter by brand, compute profit (sales minus cost of goods), and use the FLATTEN function to explore JSON food-item health metrics.
- **Big Data in Snowflake** [19:57] — A sample table (store_sales) contains 288 billion records. Traditional OLTP databases are not designed for this workload; Snowflake's architecture handles it efficiently.
- **Key Features: Cloud Native & Separated Compute/Storage** [20:45] — Snowflake is fully managed (no cluster maintenance) and runs on AWS, Azure, and GCP with the same experience. Compute and storage are decoupled, so each can scale independently.
- **Key Features: Optimization, Semi-Structured Data & More** [22:13] — Automatic performance optimization uses micropartitioning, pruning, and caching. It supports JSON via the VARIANT type and FLATTEN. Other features include time travel, zero-copy cloning, built-in security (RBAC, MFA, encryption, row/column access policies), Snowpark for Python/Java, Snowpipe for continuous ingestion, and pay-as-you-go billing.
- **Snowflake vs Traditional Data Warehouses** [26:43] — Traditional warehouses are on-prem, tightly coupled, require manual tuning, support mostly structured data, and have fixed hardware/license costs. Snowflake is cloud-native, separates compute and storage, auto-optimizes, supports semi-structured data, requires little maintenance, and offers flexible pay-per-use pricing.

### Conclusion

Snowflake is a cloud-native, fully managed data warehouse that separates compute and storage, handles massive analytical workloads, and offers modern features like semi-structured data support, time travel, and Snowpark. The tutorial provides a hands-on introduction, from account setup to running queries, and positions Snowflake as a modern alternative to traditional data warehouses.

## Transcript

understand what exactly is Snowflake, how is it different from traditional data warehouses. We will also do some hands-on practice and understand the key features and benefit. In year [snorts] 2020, one company IPOed in New York
2020, one company IPOed in New York stock market raising $3.36 billion. This stock market raising $3.36 billion. This made them a biggest software IPO in the world. That company was Snowflake. Right now 751 of Forbes global 2,000
companies are using Snowflake. So you can already see it's very very popular and if you go to any job portal you will find tons of jobs. See here on no right now I'm finding 34,000 jobs. So what exactly is Snowflake? See it's a data
warehouse. So you need to know what exactly is data warehouse and for that I have a separate video but I will quickly go over the architecture. So let's say there is a company something like Amazon or Blinket they
will have relational databases. So let's say they have some orders table say they have some orders table inventory table and this can be Oracle inventory table and this can be Oracle or Microsoft SQL server etc. They will
also have unstructured data right they will have PDF files log files some JSON objects and so on. Now if you want to do data analytics or AI, you can't do it
data analytics or AI, you can't do it directly on this uh raw data sources explained in that video. So what companies do is they will extract all this data and put them into a data lake. Data lake can be simple Amazon S3 object
storage or Azure blob storage etc. It's like a dump ground where you dump all like a dump ground where you dump all the data as it is and then you will do transform and load into your data warehouse. So here you will perform
certain transformation, you will do data cleaning. Let's say if you have business have multiple currencies. Now when you want to do the analytics, you want to bring them onto the same currency. So you might do a currency conversion, you
will do some aggregation, those kind of things happens in transform step and then you load it into a data warehouse. Now this data warehouse is your analytics and AI ready data. So here you can hook your PowerBI dashboard, your
Jupyter notebook, your AI program etc. And snowflake is this data warehouse. So there are many data warehouse uh red shift terata and so on. And Snowflake is
one of the data warehouse. It is a cloudbased SQL data warehouse. So first of all, Snowflake you can't install on your local computer. It is cloudnative your local computer. It is cloudnative and mainly you write SQL queries when
you're working with this Snowflake. So what are the differences between RDBMS and data warehouses? Well, primary use for RDBMS such as Oracle, MySQL, etc. is OLTP. You know, it's a online transaction processing system. Whereas
this is OLAP analytics. So you will use Snowflake for data warehouse. You will not use Snowflake to store your payment transactions and etc. Right? That's a transactional uh data use case. The typical workload for OLTP is many small
readrs, high concurrency transactions like orders, payment. So when you put an order in Amazon or Flipkart etc. that order information your payment of RDBMS databases whereas uh for data warehouse
the typical workload is few very large queries scanning billions of rows in data warehouse you will have billions of rows and you will have large queries you know when you're performing analytics or when you're doing an AI you are
performing large queries which will typically take time and then RDBMS MS is as SAS in one of the cloud providers. Data warehouse. When you talk about data warehouse, especially Snowflake, it is fully managed. Okay. So, whatever we're
There are other data warehouses which are not even fully managed. But Snowflake is fully managed SAS platform. So, they started Snowflake in 2012 and between 12 to 15 uh it became generally
available as a cloudnative data warehouse. Right? So that's a time warehouse. Right? So that's a time history. In 2020 they went IPO right the biggest IPO uh and they came up with this data cloud vision and launched
this data cloud vision and launched snowark. So up till here it was all SQL. So if you're working in snowflake you'll be mainly writing SQL queries but they from data bricks and so on they realize they need to give support to other
programming language. So with snowark they started supporting Python, Java and they started supporting Python, Java and so on and today they call themselves a AI data cloud. So if you go to their website they will say okay Snowflake is
all about analytics, AI, data engineering, apps and collaboration engineering, apps and collaboration power them all in AI data cloud. Okay so they are attempting or they are going into the direction of becoming unified
analytics platform. Unified analytics platform means you can do your data engineering, data analytics, AI all in one place just like data bricks. In data bricks you can do all of this in one place. So snowflake is a cloud-based AI
data platform that combines data warehousing, data engineering, machine learning, AI application development, data sharing, all of that in a unified environment. Now I can go over all kind of theory but what I would like to do
instead is get you started with snowflake free account. All right. So go to snowflake.com on their website and click on start for free. Once again it's a cloudbased environment. So you can't download any
software here. So you have to go to their website and register here. So I'm using some learning email ID for registration. Just click on continue. registration. Just click on continue. company you can give any company name
uh data engineer and doesn't matter select your region you can just say Asia wherever you are just can just say Asia wherever you are just select that region get started
you work with Python you You can select this or you can just skip this survey and it will send you an email. I got this email. I will click on that to activate my account.
to activate my account. Okay. Give your password etc and click on get started. So it will take few minutes to uh prepare the environment and then you will get this uh screen. Okay. So what I would like to do is I
would like to explore a sample data set. So just click on start here. All right. So just click on start here. All right. So it took some time to load that thing. But basically we are done uh with the registration. You get some $400 free
credits and 30 days free trial. Okay. So we will be doing our learning with that free trial. In this lecture, we are going to go through a sample use case for this tasty bit by bites fictitious global food truck chain. Snowflake
provides this data set for our learning. Here we will load the data from AWS S3 and then we'll run some SQL queries. Okay. So whatever you are seeing here,
this is SQL code, right? And this thing here is called worksheet. SQL worksheet. here is called worksheet. SQL worksheet. It's like Jupyter notebook for SQL where you can move your mouse cursor here and you can uh run this this particular set.
You know in Jupyter notebook you can run the code line by line. So it's the same thing but it is for SQL. Now let's go through this useful SQL worksheet tips. So before you work on a worksheet it has
a context which means you need to set a role warehouse database and schema. So role warehouse database and schema. So for our worksheet the role is snowflake learning role. See it is loading the roles and warehouses. So it has already
created that role as part of the default installation. installation. And then for warehouse we have this snowflake learning wh means warehouse. Okay. So see these are the roles. So we
are using the learning role and then we are using this particular warehouse. Okay. and database and schema. So if you go to databases, you have all these learning databases. So if you click on it, you will find learning databases.
it, you will find learning databases. You will also find uh here the sample data and so on. All right. So if you look at this visual, you will get more idea. But here also see you will see all this sample data. Let's go to next. And
you can run a single query or multiple queries. So here when you move your mouse cursor here click on this or control enter you will run a single query and if you say run all it will run everything. Okay then you can explore
query results and performance try using the object explorer. Okay. All try using the object explorer. Okay. All right. Now let's get started with this worksheet. So here what we are doing is essentially we are
what we are doing is essentially we are using a role. So any SQL workload that you're running will have some role right? You are a data analyst, data engineer, what kind of user you are. So this code will be doing
that. So move your mouse cursor here and press control enter if you're using Windows. If you're using Mac, there will be different command. Or you can place mouse cursor here and simply click on this button. If you say run all, it will
run all the code. But I I don't want you to run all the code. Just run first single line. The UI is initializing. It's going to take some time. You will move to the second line which is use warehouse snowflake learning warehouse.
warehouse snowflake learning warehouse. Okay. So we are having this uh learning uh environment right now. So we are using that learning warehouse. Okay. So see statement executed successfully. So we are using snowflake learning role.
Here it will show you all the query statistics like this query take took 35 millisecond uh 35 millisecond or whatever the time period is and initially it will take more time to run those queries. Then you come here again
run that particular cell. So you are now using snowflake warehouse. Okay, then go here. Control enter again. There you are uh using this database
called snowflake learning DB. Now if you go to database here, see snowflake go to database here, see snowflake learning DB. Now here you have public schema, you have information schema and you have my username is doublep. So load
you have my username is doublep. So load sample data. Okay. So see this is load sample data. So the thing that you are seeing here this snowflake learning DB seeing here this snowflake learning DB is a database. Okay. And underneath what
you have is this schema. Okay. So you have information schema and you have have information schema and you have this public one. Now under this double p whatever you don't have any objects. All right. So let's start. All right. So
let's try creating that object. So what we are going to say is set schema name is concrete current user which will be doublep see doublep
doublep see doublep load sample this okay so let's do control enter and we are saying use schema identifier schema name so my schema name will be the world p lord whatever and by default now whatever
operation we do or later on it is going to use this particular schema. Now here we are going to create a table. So the table is for a menu. See you have a food
truck chain and food truck will have a menu of your food items [snorts] and menu of your food items [snorts] and that will have menu ID, menu type ID, truck brand name, item category and so on. Okay. So this is a simple SQL
statement. And by the way, this SQL that you are running here is snowflakes own SQL dialect. There is a an SQL, right? The standard SQL that you pretty much
use everywhere and all these dialects. So for example, when you use SQL in snowflake there will be mostly it will be you'll be familiar with all the commands but some commands will vary. So snowflake folks have done their own uh
customization here. I was getting some error at this line 31 and when I reran it it just worked. Okay. So I have run all the way till this where you are creating a menu and when you click on this particular query ID you will see
the details of that query such as okay how much time it took start time end time and so on. Now with this this table should be created. So let's refresh this should be created. So let's refresh this and under this doublep load schema we
should see that table. See now you have a menu table. Initially it was like no objects and when you click on it you will see the columns etc. Somehow this snowflake instance is slow for me. Uh but when I was using it
yesterday it was pretty fast. So I don't know what's going on. But anyways we see this table now. Now when you say select star from menu control enter obviously star from menu control enter obviously there are no menu items. Okay. Now
the result will come up and I will show you that there are no menu items. Now what we are doing here is snowflake folks have put this S3 bucket with some sample data. We are going to fetch the sample data from S3 into
snowflake environment. And here we are using this command create or replace using this command create or replace stage blob stage.
documentation on this. By the way, there are two kind of stages internal and So this is the external stage that we are creating. And stage is more like a are creating. And stage is more like a pointer. Okay. So this blob stage now
will point to this directory. We have not got the data yet. Okay. So all we are doing is creating a reference link. We are saying that okay blob stage means this particular S3 location and whatever objects you have and the file format is
CSV. So there is there are CSV files uh in here and then let me just uh run this in here and then let me just uh run this thing. So click on this and execute.
And meanwhile this see menu table it's showing the schema of it. So menu ID is a number menu type ID menu type. Okay. So these are all the categories you can So these are all the categories you can go through it. It's a menu table. Now I
go through it. It's a menu table. Now I created this particular stage here and when I executed see here I created a stage right and see here I created a stage right and when I executed list which means list uh
the files which are present it shows me this okay so it has this one file which this okay so it has this one file which is CSV.gz GZ file and that is a zip file containing my menu items. Okay. Now here I will run this copy into menu command.
So what this command will do is see it will say copy into menu. Copy into menu means into this menu table. Right now this table is empty. Okay. So when we ran select star command we did not see the results. But if you have seen the
the results. But if you have seen the results it is all empty. And now we are results it is all empty. And now we are saying that copy into menu from blob stage. Blob stage is what? Blob stage is this S3 bucket. Okay, there we have row
pose and menu. Okay, so whatever data you have there, just copy that into table. So here we just created a reference called blob stage which is pointing to this S3 bucket. And here we are actually copying the data. So once
are actually copying the data. So once the data is copied now let's run this command select count star and there you go it says you have 100 rows. Now select top 10. So when you say select top 10 star it will show you the top 10 rows
from that menu table. Okay. So in this menu table now we have data from S3 and now we are saying okay show me the first 10 rows in this table. All right.
10 rows in this table. All right. See these are the menu items. So menu ID you have truck brand name you have the actual menu manual item which is lemonade beverage call option cost of goods sales price and then there is a
JSON field also. So menu item health metrics that's a JSON field and each JSON object looks something like this. So it has this key called menu item health matrix which has ingredients all these ingredients. Okay. So now here
what menu items does the freezing point brand sell. So when you run this it will show you that I'm not going to run that. You can run it because it's just taking some time. Then when you run this
particular command it will uh show you the profit. So profit is sales minus cost of goods. Right? So it's a simple SQL query folks. Okay that's what it will run. And this particular query is probing this particular JSON field and
probing this particular JSON field and it is showing you the mango sticky rice ingredient. So here we are saying that okay when menu item is mango sticky rice okay when menu item is mango sticky rice and truck brand is this go to
menu item health matrix. See menu item health matrix here and show me the ingredients. So whatever object you get dot value ingredients flatten that as an
array. Okay. So see if you have mango sticky rice. Let's run this command and see. All right. Eventually you got all these
ingredients. So to summarize in this particular lecture we loaded data from S3 and then we ran bunch of SQL queries. Okay. So here in this particular table
we loaded data and we ran bunch of SQL queries. Now folks experience-wise this looks more like you have my SQL. My SQL is a relational database right and you have bunch of databases and here you run all these queries right? So that's what
it looks like. See here on the left hand side you have databases here you are running SQL queries and you see the result. But the difference is this is OLTP it is transactional data here this is OLAP. it is analytical processing and
if you look at some of the databases here they're pretty big so see I'm in sample data and this SF00 TCL there is this table called store sales so I created a new SQL worksheet okay by clicking on this then I set the context
role warehouse database and then when I say select count star from store sales check how many records it has pause this video and you tell me how many records video and you tell me how many records is is well this is 288
billion rackovers. So this is what is big data and that's where snowflake shines in. My SQL, Oracle etc. OLTP databases are not designed to handle this kind of workload. Okay. So folks that is the beauty of snowflake and I
hope this tutorial you got started with some basic commands and you got some basic familiarity with snowflake environment. Let's discuss some of the key features and benefits of Snowflake. The first one, it is cloud native, fully
managed. You log into weather website and everything is given to you as a managed service. You don't have to do any kind of management of creating clusters, maintenance, installation and so on. This lets you focus on your
business problem and you can avoid wasting time on the infrastructure details. It also runs on AWS Azure all those majority cloud providers with the same experience. The second benefit is it has separation of compute and
storage. So in traditional systems compute and storage were tightly coupled. Okay. And sometimes what happens is let's say you don't have too happens is let's say you don't have too much of uh data usage storage usage but
you are using compute very often or you have a high compute usage in that case you have to scale both of it because these two things are tightly coupled in traditional systems but in snowflake they are independent you can scale
individually let's say you want more compute but less storage or let's say you want more storage and less compute you can configure it very easily. In you can configure it very easily. In snowflakeh example that we saw in the
one of the past lectures, we saw this uh virtual warehouse, right? So virtual virtual warehouse, right? So virtual warehouse enables you for that. Okay. So warehouses and these are virtual warehouses. Through virtual warehouse
you can you know get a scalable compute. The next benefit is automatic performance optimization. So it has bunch of
features such as micropartitioning, pruning, caching and through all these optimization you get very high performance when you are doing big data analytics. The next one is support for structured and semiructured data. So it
obviously works with the structured data such as tables and columns but it also works with semistructured data. For example, you can work with JSON. Okay, you can use variant data type to store and query semistructured data. Now, in
the SQL use case that we saw before, we saw the use of flatten. You know, we had saw the use of flatten. You know, we had this JSON column and that has some JSON data. So, JSON is a semiructured data, but you can query on that semiructured
data. You can use the functions like flatten for doing that. The next one is time travel and zero copy cloning. uh if you have attended code basics data bricks module then we we learned time travel. So you can do a query such as
travel. So you can do a query such as select star from sales at offset greater select star from sales at offset greater than uh equal to this one. Okay. So you can go back in time and query the data. It also has zero copy cloning which
means it can instantly create a clone of table or database without duplicating data. We'll go over this in detail in the future lectures. But these are some of the special features you get in snowflake. Build-in security and data
governance is a big one. Nowadays, all the organizations they ask for data governance features. Okay. And for that reason, majority of the popular data warehouses and data platforms provide these kind of features. And what are the
features? Well, number one is role based access control. Okay, this is something that every organization is looking for multiffactor authentication. See in in my company at lick when we uh talk to clients for projects this is the first
question they always ask okay what about data security how do I get data governance and using uh solutions like snowflake data bricks provides you all snowflake data bricks provides you all those built-in features always on uh
encryption then row and column level access policies so yes you can set the access control at even row level okay so let's say uh some people have access to the table and only few individual have access to certain rows. You can
configure your security and access control to that level at row and column level. And the next one is snow park and developer ecosystem. uh Snowflake used to be SQL heavy you know it was mainly designed for SQL
usage but later on they realized uh and maybe because of the competition with data bricks they realized that they need to add support for Python uh Java all these programming languages and they came up with this snow park through
came up with this snow park through which you can um get this Python support in snowflake or Java and other languages support you can also get continuous data injection with snowpipe So in snowpipe uh snowpipe is a feature through which
let's say you can set up your data pipeline in a way that let's say you have daily files coming in your landing folder in AWS S3 or let's say ADLS okay and through snowpipe you can monitor that directory and as soon as you get
the new file you can ingest that data into snowflake okay you can set those things using snowpipe and then fully elastic pay as you So when you're using snowflake you are fully elastic you know you are paying as
you go. So many organizations like this kind of billing policy you know they just want to pay per their usage. All right. So just to summarize these are the key features and benefits and you can just take a screenshot or just try
to summarize this uh in your own head organize your thoughts because this is that they they ask in a snowflake interview. You may be wondering what's the difference between snowflake and traditional data warehouses. See when
big data revolution came there were traditional data warehouses such as terod data and at that time in old days the approach was different these data warehouses were on prim okay there was no cloud native approach so let's
summarize the difference between traditional data warehouse and snowflake nowadays by the way terata is also available in cloud but I want to understand the evolution of this technology so traditional data
warehouses are typically onprem and VMbased they are tightly coupled in terms of compute and storage for snowflake this is the biggest feature folks okay just remember that it separates compute and storage and it is
cloud native okay so it avoids all kind of management your payment policy is kind of flexible and you're paying as you use either compute or storage the
you use either compute or storage the second one is snowflake has automatic partitioning, pruning, caching, there are bunch of features due to which it is very optimized. You know we in one of the previous lectures we saw that we had
280 billion rows and when you are running queries on that kind of data you need optimization you need speed and snowflake provides that. If you want to do performance tuning in traditional data warehouses see you have to do
optimization. If you have worked with even uh RDBMS such as Oracle, SQL server, MySQL, you know that whenever your query runs slow, you will create
some kind of index, you will do partitioning, you will have to do all this optimization manually. In Snowflake, you get it out of the box. The next one is traditional data warehouses typically support structured
data types. Okay? Whereas snowflake supports structured and semistructured meaning we just saw right in one of the previous lectures we had a JSON column and we were doing queries on JSON column. So JSON is a semi-structured
data and snowflake provides support for it. By the way nowadays these databases are evolving. So if you're using even posgress or many many popular databases they they are also adding this kind of semistructure data support. In terms of
maintenance, snowflake fully managed very less maintenance. Whereas in traditional data warehouses, you have to do manual backups, patching, tuning. If you have talk to any DBA, database administrator, you will understand, you
know, they spend so much time when I was at Bloomberg, they will have scheduled maintenance. The databases will be down on let's say weekend for few hours and the team will come you know they will work on weekend and they will do all
these activities. The next one is the cost model uh in snowflake is very flexible pay-per go whereas here is fixed hardware and license cost and the problem with this approach is like you might be renting or you might be buying
very expensive hardware and uh and licensing but if there is no usage then that money will go invest right it's like you are having some subscription service where you are paying every month some fixed amount let's say Netflix
you're paying every month, let's say $20, $30. Whereas this one is okay and and I use this approach where if I want to watch a movie, I go to YouTube, I rent a movie for three or $4, whatever that amount is, and I just see it. So
let's say if there are three months where I have not seen any movies and by the way, I don't watch many movies or shows. So if for three or four months, if I have not watched any movies, I will not be paying for that fixed monthly
subscription, right? So I'm just paying per go. So that kind of benefit you get per go. So that kind of benefit you get with snowflake and the cloud support. Um in terms of snowflake it it has multi cloud support. It runs on AWS, Azure,
GCP etc. Nowadays there are companies you know their entire ecosystem will be on Azure or AWS. So when you have this preferred cloud providers having preferred cloud providers having snowflake on that cloud provider makes
things easier and then for use cases if you have dynamic class modern analytics workloads you use snowflake otherwise you use traditional. So nowadays see companies when they are adopting data engineering practices they will go with
modern solution like snowflake or data bricks.