Snowflake Driver v1.11.0¶

Driver Version v1.11.0 Release Date 2026-06-18 Tested With Snowflake 10

This driver provides access to Snowflake , a cloud-based data warehouse platform.

Installation & Quickstart¶

The driver can be installed with dbc :

dbc install snowflake

Pre-requisites¶

Using the Snowflake driver requires a Snowflake account and authentication. See Getting Started With Snowflake for instructions.

Connecting¶

To connect, replace the Snowflake options below with the appropriate values for your situation and run the following:

from adbc_driver_manager import dbapi

conn = dbapi.connect(
    driver="snowflake",
    db_kwargs={
        "username": "USER",

        ### for username/password authentication: ###
        "adbc.snowflake.sql.auth_type": "auth_snowflake",
        "password": "PASS",

        ### for JWT authentication: ###
        #"adbc.snowflake.sql.auth_type": "auth_jwt",
        #"adbc.snowflake.sql.client_option.jwt_private_key": "/path/to/rsa_key.p8",

        "adbc.snowflake.sql.account": "ACCOUNT-IDENT",
        "adbc.snowflake.sql.db": "SNOWFLAKE_SAMPLE_DATA",
        "adbc.snowflake.sql.schema": "TPCH_SF1",
        "adbc.snowflake.sql.warehouse": "MY_WAREHOUSE",
        "adbc.snowflake.sql.role": "MY_ROLE"
    }
)

Note: The example above is for Python using the adbc-driver-manager package but the process will be similar for other driver managers. See adbc-quickstarts .

The driver supports connecting with individual options or connection strings.

Connection String Format¶

Snowflake URI syntax:

snowflake://user[:password]@host[:port]/database[/schema][?param1=value1&param2=value2]

This follows the Go Snowflake Driver Connection String format with the addition of the snowflake:// scheme.

Components:

scheme: snowflake:// (required)
user/password: (optional) For username/password authentication
host: (required) The Snowflake account identifier string (e.g., myorg-account1) OR the full hostname (e.g., private.network.com ). If a full hostname is used, the actual Snowflake account identifier must be provided separately via the account query parameter (see example 3).
port: The port is optional and defaults to 443.
database: Database name (required)
schema: Schema name (optional)
Query Parameters: Additional configuration options. For a complete list of parameters, see the Go Snowflake Driver Connection Parameters

Note

Reserved characters in URI elements must be URI-encoded. For example, @ becomes %40.

Examples:

snowflake://jane.doe:MyS3cr3t!@myorg-account1/ANALYTICS_DB/SALES_DATA?warehouse=WH_XL&role=ANALYST
snowflake://service_user@myorg-account2/RAW_DATA_LAKE?authenticator=oauth&application=ADBC_APP
snowflake://sys_admin@private.network.com:443/OPS_MONITOR/DBA?account=vpc-id-1234&insecureMode=true&client_session_keep_alive=true (Uses full hostname, requires explicit account parameter)

Feature & Type Support¶

Feature		Snowflake
Bulk Ingestion	Create	✅
	Append	✅
	Create/Append	✅
	Replace	✅
	Temporary Table	❌
	Target Catalog	✅
	Target Schema	✅
	Non-nullable fields are marked NOT NULL	❌
Catalog (GetObjects)	depth=catalogs	✅
	depth=db_schemas	✅
	depth=tables	✅
	depth=columns (all)	✅
Get Parameter Schema		❌
Get Table Schema		✅
Prepared Statements		✅
Transactions		✅

Types¶

Database to Arrow¶

Database Type	Snowflake
ARRAY	extension<arrow.json>
BIGINT	int64
BOOLEAN	bool
DATE	date32[day]
DOUBLE PRECISION	double
GEOGRAPHY (geography_output_format = EWKB)	extension<geoarrow.wkb>
GEOGRAPHY (geography_output_format = GEOJSON)	string
GEOMETRY (geometry_output_format = EWKB)	extension<geoarrow.wkb>
GEOMETRY (geometry_output_format = GEOJSON)	string [1]
INT	int64
NUMERIC	double
NUMERIC (use_high_precision = false)	int64 [2]
NUMERIC (use_high_precision = true)	decimal128
REAL	double
SMALLINT	int64
TIME	time64[ns]
TIMESTAMP	timestamp[us]
TIMESTAMP WITH TIME ZONE	timestamp[us] (with time zone)
VARBINARY	binary
VARCHAR	string

Arrow to Database¶

Arrow Type	Snowflake Type
	Bind	Ingest
	binary	BINARY
binary_view	BINARY	❌
bool	BOOLEAN
date32[day]	DATE
decimal128	NUMERIC	(NA/not tested)
double	DOUBLE PRECISION	REAL, NUMERIC, DOUBLE PRECISION
extension<geoarrow.wkb>	(NA/not tested)	GEOMETRY
fixed_size_binary	BINARY	BINARY(n)
float	REAL	(NA/not tested)
halffloat	REAL	(NA/not tested)
int16	SMALLINT	(NA/not tested)
int32	INT	(NA/not tested)
int64	BIGINT	SMALLINT, INT, BIGINT
large_binary	BINARY
large_list<value: int32>	(NA/not tested)	ARRAY
large_string	TEXT	STRING
list	(NA/not tested)	ARRAY
string	VARCHAR	STRING
string_view	VARCHAR	❌
time32[ms]	TIME	TIME(3)
time32[s]	TIME	TIME(0)
time64[ns]	TIME	TIME(9)
time64[us]	TIME	TIME(6)
timestamp[ms]	TIMESTAMP_NTZ(3)
timestamp[ms] (with time zone)	TIMESTAMP_LTZ(3)
timestamp[ns]	TIMESTAMP_NTZ(9)
timestamp[ns] (with time zone)	TIMESTAMP_LTZ(9)
timestamp[s]	TIMESTAMP_NTZ(0)
timestamp[s] (with time zone)	TIMESTAMP_LTZ(0)
timestamp[us]	TIMESTAMP_NTZ(6)
timestamp[us] (with time zone)	TIMESTAMP_LTZ(6)

Options¶

Connection Options¶

adbc.snowflake.sql.account

Type: string

The Snowflake account name.

adbc.snowflake.sql.auth_type

Values: (see table below). Default: auth_snowflake

How to authenticate to Snowflake.

Auth Method	Description
`auth_snowflake`	username/password
`auth_oauth`	OAuth
`auth_ext_browser`	use an external browser to access a FED and perform SSO auth
`auth_okta`	use a native OKTA URL to perform SSO authentication on Okta
`auth_jwt`	use a JWT
`auth_mfa`	username/password with MFA
`auth_pat`	use a programmatic access token
`auth_wif`	use Workload Identity Federation; must specify `adbc.snowflake.sql.client_option.identity_provider`

adbc.snowflake.sql.client_option.app_name

Type: string

The application name to report to Snowflake.

adbc.snowflake.sql.client_option.auth_token

Type: string

The auth token to use.

adbc.snowflake.sql.client_option.cache_mfa_token

Type: boolean

Whether to cache the MFA token in the OS credential manager.

adbc.snowflake.sql.client_option.client_timeout

Type: duration string (e.g. 300ms, 1.5s, or 1m30s)

Timeout for a network round trip plus reading the HTTP response. Uses Go’s time.ParseDuration format; negative values are treated as their absolute value.

adbc.snowflake.sql.client_option.config_file

Type: string (file path)

Path to the gosnowflake client configuration file used for “easy logging” (controls the driver’s log level and log output path).

adbc.snowflake.sql.client_option.disable_telemetry

Type: boolean. Default: false

When enabled, disables the driver’s usage telemetry by setting the CLIENT_TELEMETRY_ENABLED session parameter to false.

adbc.snowflake.sql.client_option.identity_provider

Values: AWS, AZURE, GCP, or OIDC

The Workload Identity Federation provider used to generate the identity attestation. Must be set when adbc.snowflake.sql.auth_type is auth_wif.

adbc.snowflake.sql.client_option.jwt_expire_timeout

Type: duration string (e.g. 300ms, 1.5s, or 1m30s)

How long a generated key-pair-authentication JWT remains valid. Uses Go’s time.ParseDuration format; negative values are treated as their absolute value.

adbc.snowflake.sql.client_option.jwt_private_key

Type: string (file path)

Path to a file containing the RSA private key (PKCS#1 or PKCS#8, PEM or DER) used to sign the JWT for key-pair authentication (auth_jwt).

adbc.snowflake.sql.client_option.jwt_private_key_pkcs8_password

Type: string

Passphrase used to decrypt an encrypted PKCS#8 private key supplied via adbc.snowflake.sql.client_option.jwt_private_key_pkcs8_value.

adbc.snowflake.sql.client_option.jwt_private_key_pkcs8_value

Type: string (PEM)

The RSA private key for key-pair authentication supplied inline as a PKCS#8 PEM value, instead of as a file path. If the PEM is an encrypted private key, also set adbc.snowflake.sql.client_option.jwt_private_key_pkcs8_password.

adbc.snowflake.sql.client_option.keep_session_alive

Type: boolean. Default: false

When enabled, keeps the Snowflake session alive (does not log it out) after the connection is closed.

adbc.snowflake.sql.client_option.login_timeout

Type: duration string (e.g. 300ms, 1.5s, or 1m30s)

Login retry timeout, excluding network round trip and reading the HTTP response. Uses Go’s time.ParseDuration format; negative values are treated as their absolute value.

adbc.snowflake.sql.client_option.ocsp_fail_open_mode

Type: boolean. Default: true (fail-open)

Controls OCSP certificate-revocation behavior. When enabled (fail-open), a revocation check that cannot complete (for example, an unreachable OCSP responder) is treated as a soft failure and the connection proceeds; when disabled (fail-closed), such a failure rejects the connection.

adbc.snowflake.sql.client_option.okta_url

Type: string (URL)

The native Okta URL (for example, https://<org>.okta.com) used for SSO when adbc.snowflake.sql.auth_type is auth_okta.

adbc.snowflake.sql.client_option.request_timeout

Type: duration string (e.g. 300ms, 1.5s, or 1m30s)

Request retry timeout for non-login requests, excluding network round trip and reading the HTTP response. Uses Go’s time.ParseDuration format; negative values are treated as their absolute value.

adbc.snowflake.sql.client_option.store_temp_creds

Type: string

Whether to cache the ID token in the OS credential manager.

adbc.snowflake.sql.client_option.tls_skip_verify

Type: string

(INSECURE) do not validate the server’s TLS certificate.

adbc.snowflake.sql.client_option.tracing

Type: string

(Deprecated) set the log level.

adbc.snowflake.sql.db

Type: string

The database name to connect to.

adbc.snowflake.sql.region

Type: string

The warehouse region.

adbc.snowflake.sql.role

Type: string

The role to use.

adbc.snowflake.sql.schema

Type: string

The schema to connect to.

adbc.snowflake.sql.uri.host

Type: string

The Snowflake host to connect to. Normally derived from the account identifier; set this only to override the host (for example, a private-link or proxy hostname).

adbc.snowflake.sql.uri.port

Type: int. Default: 443

The port to connect to.

adbc.snowflake.sql.uri.protocol

Values: http or https. Default: https

The protocol scheme used for connections.

adbc.snowflake.sql.warehouse

Type: string

The warehouse to connect to.

Options Affecting Queries¶

adbc.rpc.result_queue_size

Type: int. Default: 100

The max number of batches to buffer for each result stream. Batches are prefetched in parallel up to adbc.snowflake.rpc.prefetch_concurrency streams at a time.

Can be set on the statement. For example:

with conn.cursor() as cur:
    cur.adbc_statement.set_options(**{"adbc.rpc.result_queue_size": "200"})
    cur.execute("SELECT * FROM my_table")

adbc.snowflake.rpc.prefetch_concurrency

Type: int. Default: 5

The max number of result streams to fetch in parallel. Each stream buffers up to adbc.rpc.result_queue_size batches.

Can be set on the statement.

adbc.snowflake.sql.client_option.geography_output_format adbc.snowflake.sql.client_option.geometry_output_format : Values: GeoJSON, EKWB. Default: GeoJSON

The format used to return geometry/geography data. The default of GeoJSON will return these columns as string columns containing JSON data. If set to EWKB, instead geoarrow.wkb extension columns will be returned with CRS metadata in the extension type.

adbc.snowflake.sql.client_option.max_timestamp_precision

Values: nanoseconds, nanoseconds_error_on_overflow, microseconds. Default: nanoseconds

The Snowflake TIMESTAMP_LTZ, TIMESTAMP_NTZ, and TIMESTAMP_TZ types with nanosecond precision have a range greater than is possible to represent with Arrow nanosecond timestamps. This option controls what to do: nanoseconds will simply let the value overflow, nanoseconds_error_on_overflow will validate values and raise an error if they would overflow (at a performance cost), and microseconds will truncate to microseconds, which can represent the full range.

Can be set on the database.

adbc.snowflake.sql.client_option.stream_retry_enabled

Type: boolean. Default: false

Whether to buffer data read, retrying on failure, or directly yield the underlying result stream. If enabled, transient network failures will be retried up to a fixed number of attempts.

Can be set on the database, connection, and statement.

adbc.snowflake.sql.client_option.use_high_precision

Type: boolean. Default: true

For NUMBER columns , whether to read as Arrow decimals, or as Arrow int64/float64. Note that when disabled, there is risk of data being truncated.

Can be set on the database, connection, and statement.

adbc.snowflake.statement.ingest_compression_codec

Values: uncompressed, snappy, gzip, brotli, zstd, or lz4_raw (case-insensitive). Default: snappy

When ingesting, the compression codec to use for the Parquet files that are created and uploaded.

Can be set on the statement.

adbc.snowflake.statement.ingest_compression_level

Type: int. Default: the default level for the selected codec

When ingesting, the codec-specific compression level for the Parquet files that are created. The valid range depends on the codec; some codecs (such as snappy) ignore it.

Can be set on the statement.

adbc.snowflake.statement.ingest_copy_concurrency

Type: int. Default: 4

When ingesting, the max number of Parquet files to COPY INTO in parallel.

Can be set on the statement.

adbc.snowflake.statement.ingest_geo_type

Values: geography, geometry, or empty string. Default: empty string

Which Snowflake data type to use when ingesting columns of GeoArrow extension types (geoarrow.wkb, geoarrow.wkt). If empty, then it will be detected: if the SRID is 4326 and edges:spherical is set, then it will be ingested as GEOGRAPHY, else GEOMETRY. If explicitly set, the driver will use the specified type.

Can be set on the statement.

adbc.snowflake.statement.ingest_target_file_size

Type: int. Default: 10 MiB

When ingesting, the approximate target size of Parquet files to create. The actual size will tend to be slightly larger; if set to 0 there is no limit.

Can be set on the statement.

adbc.snowflake.statement.ingest_upload_concurrency

Type: int. Default: 8

When ingesting, the max number of Parquet files to upload in parallel.

Can be set on the statement.

adbc.snowflake.statement.ingest_use_vectorized_scanner

Type: boolean. Default: true

Whether to pass USE_VECTORIZED_SCANNER=TRUE when ingesting data via COPY INTO.

adbc.snowflake.statement.ingest_writer_concurrency

Type: int. Default: number of vCPUs detected

When ingesting, the max number of Parquet files to write in parallel.

Can be set on the statement.

adbc.snowflake.statement.query_tag

Type: string. Default: (unset)

A query tag to apply to queries, which can be used for monitoring.

Can be set on the statement.

Previous Versions¶

To see documentation for previous versions of this driver, see the following: