Brolly Academy | Snowflake

Stages in Snowflake

Internal vs External, User, Table and Named Stages with Examples (2026)

A practical guide to stages in Snowflake for October 2026: what a stage is, user, table and named internal stages versus external stages, how to create each one, storage integrations for Amazon S3 and Azure, the PUT, LIST, GET and REMOVE commands, file formats, directory tables, security, best practices and the errors beginners hit most often.

Stages in Snowflake explained: user, table and named internal stages vs external stages, CREATE STAGE examples, storage integrations, PUT, LIST and GET.

4 Stage Types SQL Examples Storage Integration Updated 9 Nov 2026
Share

By Brolly Academy Team · Brolly Academy · Updated 24 October 2026

Stages in Snowflake: internal and external stages explained - Brolly Academy

Quick answer

A stage in Snowflake is a named location that holds data files, either inside Snowflake (an internal stage) or in your own cloud storage such as Amazon S3, Azure or Google Cloud Storage (an external stage), so that Snowflake can load them into tables or unload table data into files. Think of it as the loading dock of your warehouse. Files arrive at the dock first, and then a COPY INTO command or a Snowpipe moves them into tables. Snowflake has three kinds of internal stage (user, table and named) plus external stages, and each has a different reference symbol and use.

  • User stage: @~, one per user, private, cannot be altered or dropped
  • Table stage: @%table_name, one per table, tied to that table only
  • Named internal stage: @stage_name, a database object you create, the most flexible choice
  • External stage: @stage_name that points to an S3, Azure or Google Cloud Storage URL, usually through a storage integration
  • File commands: PUT uploads and GET downloads (internal stages only), LIST shows files and REMOVE deletes them

Key takeaways

  1. A stage only stores files. Loading is done by COPY INTO or Snowpipe, which read from the stage.
  2. Use named internal stages for shared, repeatable loads and external stages when your files already live in your company's cloud bucket.
  3. PUT and GET work only with internal stages and cannot run from a Snowsight worksheet; use Snowflake CLI, a driver, or the Snowsight upload screen.
  4. For external stages, use a storage integration instead of typing cloud keys into the stage. Snowflake highly recommends this.
  5. Most stage errors come from four things: wrong path, missing privileges, a file format mismatch, or removing files before they finish loading.

1. Stages in Snowflake at a glance

Short answer: There are four stage types to know: user, table and named internal stages, and external stages; each is referenced with a different symbol after the @ sign.

Snowflake's guide on choosing an internal stage for local files describes the three internal types, and the CREATE STAGE reference covers named internal and external stages (Snowflake documentation, October 2026). Here is the whole picture in one table.

Stage type How you refer to it Who creates it Best for User stage @~ Snowflake, automatically for every user Your own test files that only you load Table stage @%orders Snowflake, automatically for every table Files that load into one table only Named internal stage @my_int_stage You, with CREATE STAGE Regular team loads, shared access, Snowsight uploads External stage @my_s3_stage You, with CREATE STAGE plus a URL Files already in S3, Azure or Google Cloud Storage

This is how a file normally travels through a stage (we will add a diagram here later):

  1. A source system, ETL job or person creates a file such as orders.csv.
  2. The file lands in a stage: PUT copies it into an internal stage, or the source writes it straight into the cloud bucket behind an external stage.
  3. You check the file with LIST.
  4. COPY INTO (a one-time or scheduled batch load) or Snowpipe (automatic loading as files arrive) reads the file using a file format.
  5. Rows land in a Snowflake table.
  6. After a successful load you clean up with REMOVE, or let the load command purge the file.

2. What is a stage in Snowflake?

Short answer: A stage is a storage location for data files that Snowflake can read from and write to; it is not a table, and you cannot query rows in it like a table.

If you are new to the platform, our guide to what Snowflake does explains the basics. For this post, you only need three terms:

  • Stage: a place where data files sit before they are loaded into tables, or after table data is unloaded into files.
  • COPY INTO: the SQL command that bulk loads files from a stage into a table (COPY INTO table) or writes query results out to a stage as files (COPY INTO location).
  • Snowpipe: Snowflake's serverless service that loads files automatically soon after they land in a stage. "Serverless" means Snowflake supplies the compute, not your own virtual warehouse.

Why does Snowflake need stages? Raw files from your laptop, an SFTP server or an S3 bucket cannot become table rows directly. They must sit somewhere Snowflake can reach first. If you come from Informatica or SSIS, a stage is close to the "source file directory" of an ETL job; Oracle DBAs can think of the directory SQL*Loader reads from.

Do not confuse an external stage with an external table. A stage points to files; an external table is a read-only table over files in an external stage (section 12, and our guide to external tables in Snowflake).

3. Internal vs external stages

Short answer: Internal stages keep files inside Snowflake-managed storage; external stages point to storage in your own cloud account, so the files stay under your company's control and billing.

Point Internal stage External stage Where files live Storage managed by Snowflake Your S3, Azure or Google Cloud Storage location How files get in PUT, or the Snowsight upload screen for named stages The cloud provider's own tools, apps or pipelines PUT and GET Supported Not supported; use the cloud provider's tools LIST and REMOVE Supported Supported (REMOVE needs delete rights in the cloud) Access setup Snowflake privileges only Storage integration (recommended) or credentials, plus Snowflake privileges Encryption choice SNOWFLAKE_FULL (default) or SNOWFLAKE_SSE Uses the cloud storage settings Typical use Uploads from laptops, one-off files, small teams Data lakes, landing zones, files shared with other tools

The URL prefixes an external stage accepts are s3://, s3china://, s3gov://, azure:// (which covers Blob storage, ADLS Gen2 and OneLake), gcs:// and s3compat:// for S3-compatible storage, which also needs an ENDPOINT value (Snowflake documentation, October 2026).

In most Indian IT services projects, upstream systems already drop files into the client's S3 bucket or Azure storage every night, so an external stage over that bucket avoids copying files twice. Internal stages suit files that come from a person or a small app.

Snowflake bills data transfer out of a region or cloud (egress) per TB, while data coming in is free (Snowflake documentation, October 2026). If your account runs in AWS Mumbai (ap-south-1), one of Snowflake's two India regions alongside Azure Central India (Pune), keep the bucket in the same region.

4. User stages and table stages

Short answer: Snowflake gives every user a private user stage and every table its own table stage automatically; they need no CREATE statement but have tight limits.

User stage (@~)

Every user gets one user stage. Only that user can use it. You cannot alter or drop it, and it does not accept file format options, so you must give the file format in the COPY command instead (Snowflake documentation, October 2026). It is handy for quick tests and personal files.

-- Simplified example: upload to your user stage, then list it
PUT file:///home/ravi/data/orders.csv @~/staged;
LIST @~/staged;

Table stage (@%table_name)

Every table has one table stage with the same name as the table. It is not a separate database object, so you cannot grant privileges on it, and it is not available for Iceberg tables. Listing a table stage needs a role that owns the table (Snowflake documentation, October 2026). Use it when files will only ever load into that one table.

-- Simplified example: upload to the stage of the ORDERS table
PUT file:///home/ravi/data/orders.csv @%orders;
LIST @%orders;

Both of these must run from Snowflake CLI (snow sql), SnowSQL or a driver, not from a Snowsight worksheet, because PUT cannot run there. SnowSQL still works but is a legacy client in maintenance mode; Snowflake CLI is the preferred client for new installs (Snowflake documentation, October 2026). On Windows, the local path in PUT starts with the drive letter after file://. A load that depends on files in your @~ breaks the day you go on leave, which is why Snowflake recommends named stages for shared and regular loads.

5. How to create a named internal stage

Short answer: Run CREATE STAGE with a name; optionally add an encryption type, a default file format, a directory table and a comment.

A named internal stage is a database object inside a schema, just like a table or view. You need the CREATE STAGE privilege on the schema for a permanent stage (Snowflake documentation, October 2026). This is the minimum, taken from Snowflake's docs example:

-- Simplified example: a basic named internal stage
CREATE STAGE my_int_stage
  ENCRYPTION = (TYPE = 'SNOWFLAKE_SSE');

Encryption for internal stages is either SNOWFLAKE_FULL (the default) or SNOWFLAKE_SSE. You cannot change the encryption type after you create the stage, so decide before you start loading (Snowflake documentation, October 2026).

A more realistic stage for a team in a Hyderabad GCC would attach a file format and a comment so everyone knows what it is for:

-- Simplified example: file format first, then a stage that uses it
CREATE OR REPLACE FILE FORMAT my_csv_format
  TYPE = CSV
  FIELD_DELIMITER = '|'
  SKIP_HEADER = 1;

CREATE STAGE sales_landing
  FILE_FORMAT = (FORMAT_NAME = 'my_csv_format')
  DIRECTORY = (ENABLE = TRUE)
  COMMENT = 'Daily pipe-delimited sales files from POS';

Here is what each option does:

  • FILE_FORMAT: the default format used when you load from this stage, either a named file format or inline options such as TYPE = CSV.
  • DIRECTORY = (ENABLE = TRUE): adds a directory table, a catalog of the files in the stage (section 9).
  • COMMENT: free text that describes the stage. It has no default.

Temporary stages

Add TEMPORARY (or TEMP) and the stage lasts only for your session. Be careful: when a temporary internal stage is dropped, Snowflake purges all its files whatever their load status, and you cannot recover them through Snowflake. Dropping a temporary external stage removes only the stage object; the files stay in your cloud storage (Snowflake documentation, October 2026).

6. How to create an external stage with a storage integration

Short answer: First create a storage integration that lets Snowflake reach your bucket through your cloud's identity system, then create the stage with a URL and STORAGE_INTEGRATION.

A storage integration is an account-level Snowflake object that stores a generated identity for your cloud storage, plus the list of locations it may access. Your cloud admin trusts that identity once, and after that nobody needs to paste access keys into Snowflake. Snowflake highly recommends storage integrations over the CREDENTIALS parameter, and for S3 it advises against long-term keys for security reasons (Snowflake documentation, October 2026). Only ACCOUNTADMIN or a role with the global CREATE INTEGRATION privilege can create one.

Amazon S3 flow

Snowflake's guide to configuring a storage integration for Amazon S3 sets out these steps (Snowflake documentation, October 2026):

1. Your AWS admin creates an IAM role with read (and if needed write and delete) access to the bucket path.
2. You create the storage integration with that role's ARN and the allowed locations.
3. You run DESC INTEGRATION and copy two values: STORAGE_AWS_IAM_USER_ARN and STORAGE_AWS_EXTERNAL_ID.
4. The AWS admin pastes those values into the IAM role's trust policy.
5. You grant USAGE on the integration to the role that will own the stage, then create the stage.

-- Simplified example: S3 storage integration and external stage
CREATE STORAGE INTEGRATION my_s3_int
  TYPE = EXTERNAL_STAGE
  STORAGE_PROVIDER = 'S3'
  ENABLED = TRUE
  STORAGE_AWS_ROLE_ARN = 'arn:aws:iam::<account_id>:role/<role_name>'
  STORAGE_ALLOWED_LOCATIONS = ('s3://<bucket>/<path>/');

DESC INTEGRATION my_s3_int;

CREATE STAGE my_s3_stage
  STORAGE_INTEGRATION = my_s3_int
  URL = 's3://<bucket>/<path>/'
  FILE_FORMAT = my_csv_format;

Azure flow

For Azure, the integration uses STORAGE_PROVIDER = 'AZURE', your AZURE_TENANT_ID and allowed locations in the form azure://account.blob.core.windows.net/container/path/. Snowflake's Azure guide lists these steps (Snowflake documentation, October 2026):

  1. Create the integration (one integration can authenticate to only one tenant).
  2. Run DESC STORAGE INTEGRATION, note AZURE_CONSENT_URL and AZURE_MULTI_TENANT_APP_NAME, then open the consent URL and accept.
  3. In the Azure portal, give the Snowflake service principal Storage Blob Data Reader (read only) or Storage Blob Data Contributor (read and write) on the storage account. Creating the service principal can take an hour or longer.
  4. Check the setup with SYSTEM$VALIDATE_STORAGE_INTEGRATION.
  5. Grant CREATE STAGE on the schema and USAGE on the integration to your role, then create the stage with STORAGE_INTEGRATION and an azure:// URL.

The stage URL must fall inside the locations listed in STORAGE_ALLOWED_LOCATIONS. Google Cloud Storage stages use gcs:// URLs; check Snowflake's GCS guide for those steps.

7. Uploading and managing files: PUT, LIST, GET and REMOVE

Short answer: PUT uploads local files to an internal stage, LIST shows what is in any stage, GET downloads from an internal stage, and REMOVE deletes files from any stage.

Command What it does Internal stages External stages PUT Uploads local files to a stage Yes No LIST (or LS) Lists files with size, MD5 and last modified time Yes Yes GET Downloads staged files to a local folder Yes No REMOVE (or RM) Deletes staged files Yes Yes

PUT

According to the PUT reference, the main options are (Snowflake documentation, October 2026):

  • PARALLEL: upload threads, 1 to 99, default 4.
  • AUTO_COMPRESS: default TRUE, which gzips the file during upload.
  • SOURCE_COMPRESSION: tells Snowflake how the file is already compressed.
  • OVERWRITE: default FALSE, so a file with the same name is not replaced. On Google Cloud, PUT always overwrites.

-- Simplified example: upload two ways
PUT file:///home/ravi/data/orders_2026_10.csv @sales_landing;
PUT file:///home/ravi/data/big_file.csv @sales_landing
  PARALLEL = 8 OVERWRITE = TRUE;

Uploading through Snowsight (no PUT needed)

If you do not want to install a command-line client, Snowflake's guide to staging files using Snowsight gives this route (Snowflake documentation, October 2026):

  1. Sign in to Snowsight.
  2. Go to Ingestion, then Add Data.
  3. Choose Load files into a Stage.
  4. Select one or more files.
  5. Pick the database, schema and stage, and optionally a path inside the stage.
  6. Click Upload.

This works only for named internal stages, not user or table stages, and each file can be up to 250 MB. Your role needs USAGE on the database and schema and WRITE on the stage. For bigger files, use PUT.

LIST

LIST works on all stage types. The PATTERN option takes a regular expression (Java Pattern syntax), and a path narrows the listing, which also makes it faster. The output columns are name, size (compressed bytes), md5, sha1 (not used) and last_modified (Snowflake documentation, October 2026). Note that paths are case-sensitive.

-- Simplified example: list files
LIST @sales_landing;
LIST @sales_landing PATTERN = '.*orders_2026.*';
LS @~;

GET

GET downloads files from an internal stage to a local folder. PARALLEL runs from 1 to 99 with a default of 10, and PATTERN filters the files. GET cannot download from external stages, does not keep the stage's folder structure, does not rename files, and decrypts files automatically. Like PUT, it cannot run from a Snowsight worksheet (Snowflake documentation, October 2026).

-- Simplified example: download files to a local folder
GET @sales_landing file:///tmp/download/;
GET @%orders file:///tmp/data/;

GET is useful after you unload a query result to a stage with COPY INTO location and want the files on your machine.

REMOVE

REMOVE deletes files from internal or external stages. Three rules from Snowflake's docs (October 2026):

  • Do not remove a file until its data has loaded successfully. Check with COPY_HISTORY first.
  • Add a trailing slash to a path. Without it, REMOVE also deletes objects whose names merely start with the same text, such as myobject_new when you meant myobject.
  • On external stages, Snowflake needs delete permission in your cloud account, for example s3:DeleteObject on S3 or Storage Blob Data Contributor on Azure. Files already deleted stay deleted even if the command is interrupted.

-- Simplified example: remove files
REMOVE @sales_landing/2026/10/;
RM @~ PATTERN = '.*jun.*';

8. File formats and stages

Short answer: A file format tells Snowflake how to read a staged file (CSV, JSON, Parquet and so on); you can attach it to a stage, a named file format object or the COPY command.

Snowflake supports CSV (the default), JSON and PARQUET for both loading and unloading, and AVRO, ORC and XML for loading only. The default CSV field delimiter is a comma (Snowflake documentation, October 2026). The CREATE STAGE reference also accepts TYPE = CUSTOM.

You have three places to define the format, and the more specific one wins in practice:

  1. Named file format object (CREATE FILE FORMAT): reusable across stages and loads. Best for teams.
  2. Stage default (FILE_FORMAT on the stage): every load from that stage uses it unless told otherwise.
  3. Inside the COPY command: good for one-off loads. This is the only option for user stages, which accept no file format options.

A practical tip: keep one named file format per source system (for example ff_pos_csv for core banking or POS extracts and ff_app_json for app events) and reference it from both the stage and your COPY commands. When the source team changes the delimiter, you change one object, not twenty scripts. For how JSON lands in a VARIANT column, see our post on Snowflake data types.

9. Directory tables on stages

Short answer: A directory table is a built-in catalog of the files in a stage, with each file's path, size, last modified time and file URL, that you can query with SQL.

A directory table is an implicit object layered on an internal or external stage. You add it with DIRECTORY = (ENABLE = TRUE) in CREATE STAGE or ALTER STAGE (Snowflake documentation, October 2026). It is most useful for unstructured files such as PDFs, images and scanned forms, where you need a list of files rather than rows inside them.

-- Simplified example: enable, refresh and query a directory table
ALTER STAGE sales_landing SET DIRECTORY = (ENABLE = TRUE);
ALTER STAGE sales_landing REFRESH;
SELECT * FROM DIRECTORY(@sales_landing);

The query returns the columns RELATIVE_PATH, SIZE, LAST_MODIFIED, MD5, ETAG and FILE_URL (Snowflake documentation, October 2026).

How refresh works and what it costs:

  • Manual refresh: ALTER STAGE ... REFRESH, billed as cloud services usage.
  • Automatic refresh: uses cloud event notifications (AUTO_REFRESH), billed as Snowpipe charges that grow with the number of files added.
  • Privileges: querying needs READ on an internal stage (READ or USAGE on an external one); refreshing needs WRITE (or USAGE on an external stage).

Common uses are building a list of every invoice PDF in a stage, joining file metadata with a table to make a view, and feeding a file-processing pipeline.

10. Stage security and privileges

Short answer: Internal stages use READ and WRITE privileges, external stages use USAGE, and external stages that rely on a storage integration also need USAGE on that integration for the stage owner.

Snowflake controls access with RBAC (role-based access control): privileges go to roles and roles go to users. Here are the stage privileges from Snowflake's access control docs (October 2026):

Privilege Applies to Allows READ Internal stages GET, LIST and COPY INTO table WRITE Internal stages PUT, REMOVE and COPY INTO location USAGE External stages only Using the stage in SQL; includes read and write OWNERSHIP All named stages Full control; only one role holds it

A few rules worth remembering:

  • When you grant both READ and WRITE, grant READ first or in the same statement. When revoking both, revoke WRITE first or together.
  • Table stages and user stages are not first-class objects, so they have no grantable privileges. Listing a table stage needs OWNERSHIP on the table; listing your own user stage needs no privilege.
  • For an external stage with a storage integration, the role that owns the stage needs USAGE on the integration. The role running LIST or REMOVE does not.
  • Every stage operation also needs at least one privilege on the parent database and schema.

On the security side, keep STORAGE_ALLOWED_LOCATIONS narrow, use Storage Blob Data Reader on Azure if Snowflake never needs to write, and if you must use CREDENTIALS for a test, put the URL in single quotes so the credentials do not show in plain text in query history (Snowflake documentation, October 2026).

11. How stages feed COPY INTO, Snowpipe, external tables and Openflow

Short answer: A stage is the starting point for almost every file-based load: COPY INTO and Snowpipe read from stages, external tables query files in external stages, and unloading writes back to stages.

The next step after staging is usually a bulk load. In its simplest form it is two lines:

-- Simplified example: the next step after staging
COPY INTO orders FROM @sales_landing
  FILE_FORMAT = (FORMAT_NAME = 'my_csv_format');

Options such as ON_ERROR, PURGE and FORCE, load history and unloading are covered step by step in our COPY INTO Snowflake lab.

Here is how each loading path uses stages:

  • COPY INTO table: reads files from any stage type. Best for scheduled batch loads that you run on a warehouse.
  • Snowpipe: wraps a COPY statement in a pipe and loads files automatically as they land in cloud storage, usually an external stage with event notifications. Our Snowpipe auto-ingest lab walks through it. Snowpipe Streaming is different: it loads rows without staging files at all (Snowflake documentation, October 2026).
  • External tables: read-only tables over files in an external stage, so you can query without loading. See external tables in Snowflake.
  • COPY INTO location: unloads query results as files into a stage, which you then download with GET or read from your bucket.
  • Openflow: Snowflake's NiFi-based integration service, which can pull data from sources such as SharePoint, Kafka or databases. Read our Snowflake Openflow guide to see where it fits beside stages and Snowpipe.

After loading, streams and tasks often take over for change processing and scheduling. For an end-to-end walkthrough that creates a stage, loads a CSV and queries it, follow our hands-on Snowflake tutorial.

12. Which stage when: an interview-style decision table

Short answer: Pick the stage by where the files come from, who loads them and whether they must stay in your cloud account.

"Which stage would you use?" is a common scenario question in Snowflake interviews. This table gives you a crisp answer.

Situation Best choice Why You test a CSV on your laptop for your own practice table User stage (@~) Zero setup and private to you A file will only ever load into one table, loaded by its owner Table stage (@%table) Tied to that table, no extra object A team loads daily files, several roles need access Named internal stage Grantable privileges, default file format, Snowsight upload Upstream systems already drop files into S3 or Azure External stage with storage integration No double copy, files stay under your control Files must auto-load within minutes of arrival External stage plus Snowpipe Event notifications trigger the pipe You want to query lake files without loading them External stage plus external table Read-only access in place You need a list of PDFs or images with URLs Any named stage with a directory table File metadata you can query

External stage vs external table

  • External stage: a pointer to a cloud storage location, used for loading, unloading and file commands. You cannot query rows through it like a table.
  • External table: a read-only table built on files in an external stage. It has a VALUE column plus metadata columns such as METADATA$FILENAME, and you can query, join and build views on it, but not run DML (Snowflake documentation, October 2026).

So every external table needs an external stage, but an external stage does not need an external table.

Stage vs storage integration

A stage says where the files are. A storage integration says how Snowflake is allowed to reach them. One storage integration can serve many external stages, as long as each stage URL sits inside its allowed locations.

13. What stages cost

Short answer: The costs to watch are files kept in internal stages, which use Snowflake storage, plus the compute for loading, directory table refreshes and cross-region transfers.

Snowflake's total cost is compute plus storage plus data transfer (Snowflake documentation, October 2026). For stages, that breaks down like this:

  • Internal stage files: count as Snowflake storage, billed at a flat rate per TB per month based on average daily bytes. Old files you forget to remove keep costing money.
  • External stage files: live in your cloud account, so your cloud provider bills the storage, not Snowflake.
  • Loading: COPY INTO uses your virtual warehouse, which bills per second with a 60-second minimum each time it starts. Snowpipe uses Snowflake-managed serverless compute.
  • Directory tables: automatic refresh is billed as Snowpipe charges; manual refresh counts as cloud services usage.
  • Data transfer: ingress is free; egress across regions or clouds is charged per TB.

Prices vary by cloud, region and edition, so check your own contract or Snowflake's consumption table rather than relying on blog numbers. The cheapest habit is simple: remove or purge files once they are safely loaded.

14. Best practices for stages

Short answer: Use named stages, organise files by path, attach file formats, use storage integrations, and clean up after successful loads.

A checklist we suggest for real projects:

  1. Prefer named stages over user and table stages for anything a team depends on.
  2. One stage per source or zone, such as raw_pos_stage and raw_app_stage, each with a COMMENT.
  3. Use date-based paths such as /2026/10/25/ so LIST, COPY and REMOVE can target a day's files.
  4. Attach a named file format to each stage so loads behave the same every time.
  5. Use storage integrations with narrow allowed locations for every external stage.
  6. Grant least privilege: WRITE only to loading roles; USAGE on external stages only to roles that need it.
  7. Check before you load: run LIST with a PATTERN to confirm the files and sizes.
  8. Clean up: REMOVE files after COPY_HISTORY confirms the load, and always end paths with a slash.
  9. Choose encryption at creation for internal stages, because it cannot be changed later.

15. Common stage errors and how to fix them

Short answer: Most stage problems are path, privilege, client or file format issues, and each has a quick check.

Error or symptom Likely cause Fix PUT or GET fails in a Snowsight worksheet These commands cannot run there Use Snowflake CLI (snow sql), a driver, or the Snowsight upload screen PUT to an external stage fails PUT works only with internal stages Upload with the cloud provider's tools LIST returns nothing Wrong path case, wrong stage or wrong pattern Paths are case-sensitive; list the stage root first Insufficient privileges on a table stage Your role does not own the table Use the owning role or switch to a named stage Errors on a new S3 or Azure stage URL outside STORAGE_ALLOWED_LOCATIONS, or trust not set Match the URL to allowed locations; finish the IAM trust or Azure consent step Access denied on an external stage Stage owner lacks USAGE on the integration, or the cloud role lacks rights Grant USAGE on the integration; check the IAM role or Azure role assignment Same file not replaced on re-upload OVERWRITE defaults to FALSE Add OVERWRITE = TRUE when you really mean to replace REMOVE deleted more files than expected Path without a trailing slash matched other prefixes Always end folder paths with / Snowsight upload refused File over 250 MB, or target is a user or table stage Use PUT, or upload to a named internal stage COPY reads columns wrongly File format does not match the file Fix delimiter, header or type in the named file format

16. How Brolly Academy helps

Short answer: Brolly Academy's 45-day Snowflake course covers file formats, internal and external stages, COPY INTO, PUT and GET, and Snowpipe as part of a 15-module syllabus with hands-on projects.

Stages are not a side topic. In the job descriptions we read on 25 October 2026, an Infosys Snowflake admin role in Hyderabad listed Snowpipe, stages and data sharing among its duties, and roles at Capgemini, TCS and DTCC named Snowpipe or external tables. SnowPro Core also gives 18% of its weight to Data Loading, Unloading and Connectivity (learn.snowflake.com, October 2026).

Our Snowflake training in Hyderabad covers these skills in two modules. Module 6, data loading and unloading, covers file formats, internal and external stages, COPY INTO, loading from AWS, Azure and Google Cloud Storage, and unloading. Module 7 covers Snowpipe, including PUT and GET, bulk loading and continuous loading. Module 9 adds JSON and semi-structured data with VARIANT, file formats and stages.

  • Duration and modes: 45 days, as classroom, live online or a self-paced video plan
  • Fees: Video ₹8,999, Online ₹17,999, Classroom ₹22,999 (EMI available)
  • Trainer: Mr. Bala Krishna, 13+ years in data warehousing, SQL and Snowflake
  • Outcome: 15 modules, 6 projects and a Brolly Academy course completion certificate; SnowPro exam fees are separate and paid to Snowflake

Load your first file through a stage with a trainer beside you

Reading about PUT, LIST and storage integrations is the easy part; making an S3 stage work the first time is where most learners get stuck. Join a free demo class at Brolly Academy, see how our 45-day Snowflake course teaches stages, COPY INTO and Snowpipe hands-on, and decide whether classroom, live online or the video plan suits you.

Frequently asked questions

What is a stage in Snowflake?

A stage in Snowflake is a location that holds data files so Snowflake can load them into tables or unload table data into files. It can be internal, inside Snowflake-managed storage, or external, pointing to Amazon S3, Azure or Google Cloud Storage. A stage only stores files. Commands such as COPY INTO and services such as Snowpipe read from the stage and move the data into tables.

What are the types of stages in Snowflake?

Snowflake has four stage types you need to know. Internal stages come in three kinds: the user stage referenced as @~, the table stage referenced as @% plus the table name, and the named internal stage you create with CREATE STAGE. The fourth type is the external stage, which points to a cloud storage URL in S3, Azure or Google Cloud Storage, usually through a storage integration.

What is the difference between an internal and an external stage?

An internal stage keeps files in storage that Snowflake manages, while an external stage points to storage in your own cloud account. You upload to internal stages with PUT or the Snowsight upload screen, and download with GET. External stages do not support PUT or GET, so files arrive through the cloud provider's own tools. LIST and REMOVE work on both types.

What is a user stage in Snowflake?

A user stage is a private internal stage that Snowflake creates automatically for every user, referenced as @~. Only that user can use it. You cannot alter or drop it, and it accepts no file format options, so you give the format in the COPY command. It suits quick personal tests, but team loads should not depend on it because teammates cannot see its files.

What is a table stage in Snowflake?

A table stage is an internal stage that Snowflake creates for every table, referenced as @% followed by the table name. It is not a separate database object, so you cannot grant privileges on it, and it is not available for Iceberg tables. Listing it needs a role that owns the table. Use it only when files will load into that one table.

What is a named stage in Snowflake?

A named stage is a stage you create yourself with CREATE STAGE as an object inside a schema. It can be internal or external. Named stages are the most flexible choice because you can grant privileges on them, attach a default file format, add a directory table and a comment, and upload to them through Snowsight. Snowflake recommends them for shared and regular loads.

How do I create a stage in Snowflake?

You create a named internal stage with CREATE STAGE followed by a name, and you can add options such as ENCRYPTION, FILE_FORMAT, DIRECTORY and COMMENT. For an external stage, you add a URL such as an s3:// or azure:// path and a STORAGE_INTEGRATION. Creating a permanent stage needs the CREATE STAGE privilege on the schema, and using an integration needs USAGE on it.

How do I upload a file to a Snowflake stage?

You upload a local file to an internal stage with the PUT command, run from Snowflake CLI, SnowSQL or a driver. PUT compresses files with gzip by default and uses four upload threads unless you set PARALLEL. For named internal stages, you can also use the Snowsight upload screen under Ingestion and Add Data, which accepts files up to 250 MB each.

Can I run PUT from a Snowsight worksheet?

No, PUT and GET cannot run from a Snowsight worksheet. Run them from Snowflake CLI with snow sql, from SnowSQL, or through a driver. If you want a browser-only route, the Snowsight upload screen lets you add files to a named internal stage without PUT. That screen does not work for user stages, table stages or external stages, and caps each file at 250 MB.

How do I list files in a Snowflake stage?

You list files with the LIST command, or its short form LS, followed by the stage reference and an optional path. Add PATTERN with a regular expression to filter file names. The output shows each file's name, compressed size, MD5 value and last modified time. Paths are case-sensitive, and giving a path narrows the listing and makes it faster.

How do I download files from a Snowflake stage?

You download files with the GET command, which copies files from an internal stage to a local folder. It uses ten download threads by default and accepts a PATTERN filter. GET cannot download from external stages, does not keep the stage's folder structure, and does not rename files. Like PUT, it must run from Snowflake CLI, SnowSQL or a driver, not a Snowsight worksheet.

How do I delete files from a Snowflake stage?

You delete staged files with the REMOVE command, or RM, on internal or external stages. Check COPY_HISTORY first so you never remove a file that has not finished loading. End folder paths with a slash, because without it REMOVE also deletes objects whose names start with the same text. On external stages, Snowflake needs delete permission in your cloud account.

What is a storage integration in Snowflake?

A storage integration is an account-level Snowflake object that stores a generated cloud identity and a list of allowed storage locations. Your cloud admin trusts that identity once, through an IAM role trust policy on AWS or a consent and role assignment on Azure. External stages then use the integration instead of access keys. Only ACCOUNTADMIN or a role with CREATE INTEGRATION can create one.

Why use a storage integration instead of credentials?

You should use a storage integration because it avoids typing cloud access keys into stage definitions or load commands. Snowflake highly recommends integrations over the CREDENTIALS parameter and advises against long-term S3 keys for security reasons. One integration can also serve many stages, and its allowed locations list limits which buckets and paths Snowflake can ever reach.

What is the difference between an external stage and an external table?

An external stage is a pointer to files in cloud storage, used for loading, unloading and file commands. An external table is a read-only table built on files in an external stage, with a VALUE column and metadata columns such as METADATA$FILENAME. You can query and join an external table, but you cannot run DML on it. Every external table needs an external stage.

Is there a file size limit for Snowflake stages?

The clearest documented limit is for the Snowsight upload screen, which accepts files up to 250 MB each and only for named internal stages. For larger files, use the PUT command from Snowflake CLI or a driver, where you can raise PARALLEL to speed up the upload. For external stages, file sizes follow your cloud storage, because files arrive through the provider's tools.

Do internal stages cost money in Snowflake?

Yes, files kept in internal stages count as Snowflake storage, billed at a flat rate per TB per month, so forgotten files keep adding to the bill. Loading uses warehouse compute or serverless Snowpipe compute, and directory table auto-refresh is billed as Snowpipe. Prices vary by cloud, region and edition, so check your own contract.

What file formats can a Snowflake stage use?

A Snowflake stage can work with CSV, JSON, PARQUET, AVRO, ORC and XML files. CSV, JSON and PARQUET work for both loading and unloading, while AVRO, ORC and XML are for loading only. CSV is the default, with a comma as the default delimiter. You can attach a format to the stage, use a named file format object, or set it in COPY.

What is a directory table on a stage?

A directory table is a built-in catalog of the files in an internal or external stage. You enable it with DIRECTORY equals ENABLE TRUE and query it with SELECT from DIRECTORY on the stage. It returns each file's relative path, size, last modified time, MD5, ETag and file URL. Manual refresh uses ALTER STAGE REFRESH, while automatic refresh is billed as Snowpipe.

Can I unload data from Snowflake to a stage?

Yes, you can unload query results into files on a stage with COPY INTO location. Your role needs WRITE on an internal stage, or USAGE on an external stage, to do this. After unloading to an internal stage, you can download the files with GET. Unloading supports CSV, JSON and PARQUET formats, and the COPY INTO guide linked above covers the options in detail.

What privileges do I need to use a stage?

On a named internal stage, READ allows GET, LIST and loading with COPY, while WRITE allows PUT, REMOVE and unloading. External stages use USAGE, which covers both read and write. You also need at least one privilege on the parent database and schema. For external stages with a storage integration, the role that owns the stage needs USAGE on the integration.

What is a temporary stage in Snowflake?

A temporary stage is a stage created with the TEMPORARY or TEMP keyword that lasts only for the session that created it. Be careful with temporary internal stages: when one is dropped, Snowflake purges all its files whatever their load status, and they cannot be recovered through Snowflake. Dropping a temporary external stage removes only the stage, and the files stay in cloud storage.

How do stages work with Snowpipe?

Snowpipe wraps a COPY statement in a pipe and loads files automatically as they land in cloud storage, usually behind an external stage with event notifications. The stage tells Snowpipe where the files are and which file format to use. Snowpipe uses serverless compute managed by Snowflake. Snowpipe Streaming is different because it loads rows directly without staging files at all.

Are stages asked in Snowflake interviews and the SnowPro Core exam?

Yes, stages come up often. SnowPro Core gives 18% of its weight to Data Loading, Unloading and Connectivity, according to learn.snowflake.com in October 2026. Job descriptions we read that month, including an Infosys Snowflake admin role in Hyderabad, named stages and Snowpipe among the duties. Practise explaining the stage types, internal versus external, and external stage versus external table.

Where can I learn Snowflake stages hands-on in Hyderabad?

Brolly Academy's 45-day Snowflake course in Hyderabad teaches stages hands-on. Module 6 covers file formats, internal and external stages, COPY INTO and loading from AWS, Azure and Google Cloud Storage, and Module 7 covers Snowpipe with PUT and GET. The course runs as classroom, live online or a video plan, with fees from ₹8,999, and SnowPro exam fees are separate.

Learn with Brolly Academy

Review the syllabus, practical projects and batch timings, and book a free demo before you enrol.

Related Articles

Snowflake Tutorial

Hands-on guide for beginners.

Read Article

Snowpipe

Auto-ingest from S3 step by step.

Read Article

COPY INTO Snowflake

CSV loading with validation.

Read Article

Continue Your Learning

Snowflake Training in Hyderabad

45-day course covering Snowflake architecture, warehouses, data loading, security, performance and real-time projects, in classroom, live online and recorded plans.

View Course

dbt Course

Learn dbt, the transformation tool most Snowflake job descriptions ask for, to build tested, documented data models.

View Course

Databricks Course

Add the lakehouse side of modern data engineering with Databricks, Spark and Delta Lake, often used alongside Snowflake.

View Course