Hive connector with Amazon S3#

The Hive connector can read and write tables that are stored in Amazon S3 or S3-compatible systems. This is accomplished by having a table or database location that uses an S3 prefix, rather than an HDFS prefix.

Trino uses its own S3 filesystem for the URI prefixes s3://, s3n:// and s3a://.

S3 configuration properties#

Property name


Default AWS access key to use.

Default AWS secret key to use.


IAM role to assume.


External ID for the IAM role trust policy.


The S3 storage endpoint server. This can be used to connect to an S3-compatible storage system instead of AWS. When using v4 signatures, it is recommended to set this to the AWS region-specific endpoint (e.g., http[s]://s3.<AWS-region>


Optional property to force the S3 client to connect to the specified region only.

The S3 storage class to use when writing the data. Currently only STANDARD and INTELLIGENT_TIERING storage classes are supported. Default storage class is STANDARD


Specify a different signer type for S3-compatible storage. Example: S3SignerType for v2 signer type


Specify a different signer class for S3-compatible storage.


Use path-style access for all requests to the S3-compatible storage. This is for S3-compatible storage that doesn’t support virtual-hosted-style access, defaults to false.


Local staging directory for data written to S3. This defaults to the Java temporary directory specified by the JVM system property

Pin S3 requests to the same region as the EC2 instance where Trino is running, defaults to false.


Use HTTPS to communicate with the S3 API, defaults to true.


Use S3 server-side encryption, defaults to false.


The type of key management for S3 server-side encryption. Use S3 for S3 managed keys, CUSTOMER for customer-provided keys, or KMS for KMS-managed keys. Defaults to S3.


The key when using a customer-provided key for S3 server-side encryption.


The KMS Key ID to use for S3 server-side encryption with KMS-managed keys. If not set, the default key is used.


If set, use S3 client-side encryption and use the AWS KMS to store encryption keys and use the value of this property as the KMS Key ID for newly created objects.


If set, use S3 client-side encryption and use the value of this property as the fully qualified name of a Java class which implements the AWS SDK’s EncryptionMaterialsProvider interface. If the class also implements Configurable from the Hadoop API, the Hadoop configuration will be passed in after the object has been created.


Canned ACL to use while uploading files to S3, defaults to PRIVATE. If the files are to be uploaded to an S3 bucket owned by a different AWS user, the canned ACL has to be set to one of the following: AUTHENTICATED_READ, AWS_EXEC_READ, BUCKET_OWNER_FULL_CONTROL, BUCKET_OWNER_READ, LOG_DELIVERY_WRITE, PUBLIC_READ, PUBLIC_READ_WRITE. Refer to the AWS canned ACL <>_ guide to understand each option’s definition.


Ignore Glacier objects rather than failing the query. This skips data that may be expected to be part of the table or partition. Defaults to false.


Use S3 multipart upload API to upload file in streaming way, without staging file to be created in the local file system.


The part size for S3 streaming upload. Defaults to 16MB.

Proxy host to use if connecting through a proxy.


Proxy port to use if connecting through a proxy.


Proxy protocol. HTTP or HTTPS , defaults to HTTPS.


Hosts list to access without going through the proxy.


Proxy user name to use if connecting through a proxy.


Proxy password to use if connecting through a proxy.


Whether to attempt to authenticate preemptively against proxy when using base authorization, defaults to false.


Optional override for the sts endpoint given that IAM role based authentication via sts is used.


Optional override for the sts region given that IAM role based authentication via sts is used.

Filter based on storage class of S3 object, defaults to READ_ALL.

S3 credentials#

If you are running Trino on Amazon EC2, using EMR or another facility, it is recommended that you use IAM Roles for EC2 to govern access to S3. To enable this, your EC2 instances need to be assigned an IAM Role which grants appropriate access to the data stored in the S3 bucket(s) you wish to use. It is also possible to configure an IAM role with hive.s3.iam-role that is used for accessing any S3 bucket. This is much cleaner than setting AWS access and secret keys in the and settings, and also allows EC2 to automatically rotate credentials on a regular basis without any additional work on your part.

Custom S3 credentials provider#

You can configure a custom S3 credentials provider by setting the configuration property trino.s3.credentials-provider to the fully qualified class name of a custom AWS credentials provider implementation. The property must be set in the Hadoop configuration files referenced by the hive.config.resources Hive connector property.

The class must implement the AWSCredentialsProvider interface and provide a two-argument constructor that takes a and a Hadoop org.apache.hadoop.conf.Configuration as arguments. A custom credentials provider can be used to provide temporary credentials from STS (using STSSessionCredentialsProvider), IAM role-based credentials (using STSAssumeRoleSessionCredentialsProvider), or credentials for a specific use case (e.g., bucket/user specific credentials).

S3 security mapping#

Trino supports flexible security mapping for S3, allowing for separate credentials or IAM roles for specific users or buckets/paths. The IAM role for a specific query can be selected from a list of allowed roles by providing it as an extra credential.

Each security mapping entry may specify one or more match criteria. If multiple criteria are specified, all criteria must match. Available match criteria:

  • user: Regular expression to match against username. Example: alice|bob

  • group: Regular expression to match against any of the groups that the user belongs to. Example: finance|sales

  • prefix: S3 URL prefix. It can specify an entire bucket or a path within a bucket. The URL must start with s3:// but will also match s3a or s3n. Example: s3://bucket-name/abc/xyz/

The security mapping must provide one or more configuration settings:

  • accessKey and secretKey: AWS access key and secret key. This overrides any globally configured credentials, such as access key or instance credentials.

  • iamRole: IAM role to use if no user provided role is specified as an extra credential. This overrides any globally configured IAM role. This role is allowed to be specified as an extra credential, although specifying it explicitly has no effect, as it would be used anyway.

  • roleSessionName: Optional role session name to use with iamRole. This can only be used when iamRole is specified. If roleSessionName includes the string ${USER}, then the ${USER} portion of the string will be replaced with the current session’s username. If roleSessionName is not specified, it defaults to trino-session.

  • allowedIamRoles: IAM roles that are allowed to be specified as an extra credential. This is useful because a particular AWS account may have permissions to use many roles, but a specific user should only be allowed to use a subset of those roles.

  • kmsKeyId: ID of KMS-managed key to be used for client-side encryption.

  • allowedKmsKeyIds: KMS-managed key IDs that are allowed to be specified as an extra credential. If list cotains “*”, then any key can be specified via extra credential.

The security mapping entries are processed in the order listed in the configuration JSON. More specific mappings should thus be specified before less specific mappings. For example, the mapping list might have URL prefix s3://abc/xyz/ followed by s3://abc/ to allow different configuration for a specific path within a bucket than for other paths within the bucket. You can set default configuration by not including any match criteria for the last entry in the list.

In addition to the rules above, the default mapping can contain the optional useClusterDefault boolean property with the following behavior:

  • false - (is set by default) property is ignored.

  • true - This causes the default cluster role to be used as a fallback option. It can not be used with the following configuration properties:

    • accessKey

    • secretKey

    • iamRole

    • allowedIamRoles

If no mapping entry matches and no default is configured, the access is denied.

The configuration JSON can either be retrieved from a file or REST-endpoint specified via

Example JSON configuration:

  "mappings": [
      "prefix": "s3://bucket-name/abc/",
      "iamRole": "arn:aws:iam::123456789101:role/test_path"
      "user": "bob|charlie",
      "iamRole": "arn:aws:iam::123456789101:role/test_default",
      "allowedIamRoles": [
      "prefix": "s3://special-bucket/",
      "accessKey": "AKIAxxxaccess",
      "secretKey": "iXbXxxxsecret"
      "prefix": "s3://encrypted-bucket/",
      "kmsKeyId": "kmsKey_10",
      "user": "test.*",
      "iamRole": "arn:aws:iam::123456789101:role/test_users"
      "group": "finance",
      "iamRole": "arn:aws:iam::123456789101:role/finance_users"
      "iamRole": "arn:aws:iam::123456789101:role/default"

Property name


The JSON configuration file or REST-endpoint URI containing security mappings.

A JSON pointer (RFC 6901) to mappings inside the JSON retrieved from the config file or REST-endpont. The whole document (“”) by default.

The name of the extra credential used to provide the IAM role.

The name of the extra credential used to provide the KMS-managed key ID.

How often to refresh the security mapping configuration.

The character or characters to be used in place of the colon (:) character when specifying an IAM role name as an extra credential. Any instances of this replacement value in the extra credential value will be converted to a colon. Choose a value that is not used in any of your IAM ARNs.

Tuning properties#

The following tuning properties affect the behavior of the client used by the Trino S3 filesystem when communicating with S3. Most of these parameters affect settings on the ClientConfiguration object associated with the AmazonS3Client.

Property name




Maximum number of error retries, set on the S3 client.



Maximum number of read attempts to retry.



Use exponential backoff starting at 1 second up to this maximum value when communicating with S3.

10 minutes


Maximum time to retry communicating with S3.

10 minutes


TCP connect timeout.

5 seconds


TCP connect TTL, which affects connection reusage.

Connections do not expire.


TCP socket read timeout.

5 seconds


Maximum number of simultaneous open connections to S3.



Minimum file size before multi-part upload to S3 is used.

16 MB


Minimum multi-part upload part size.

5 MB

S3 data encryption#

SEP supports reading and writing encrypted data in S3 using server-side encryption or client-side encryption.

Use server-side encryption with either S3 managed keys or customer-provided keys. Use client-side encryption with either the Amazon KMS or a software plugin to manage AES encryption keys.

Server-side encryption with S3 managed keys#

S3 server-side encryption with S3 managed keys (SSE-S3) ensures that the encryption and decryption processes are managed by the S3 infrastructure.

To enable SSE-3, set hive.s3.sse.enabled to true.


To enable S3 to administer all encryption keys, you must disable SSL to your client. To disable SSL, set hive.s3.ssl.enabled to false.

Server-side encryption with customer-provided keys#

Server-side encryption with customer-provided keys (SSE-C) allows you to to use a custom key to encrypt your data stored in S3.

To enable SSE-C, set the following configuration properties:


Client-side encryption#

With S3 client-side encryption, S3 stores encrypted data, but the encryption keys are managed outside of S3. SEP encrypts and decrypts the data. In this case, encryption keys can be managed using the AWS Key Management Service (KMS) or your own key management system.

To enable the KMS for key management, set hive.s3.kms-key-id to the UUID of a KMS key. Ensure that you grant permission to your AWS account or EC2 IAM role to use the key.

To enable a custom encryption key management system, set hive.s3.encryption-materials-provider to the fully qualified name of a class that implements the EncryptionMaterialsProvider interface from the AWS Java SDK. Ensure that this class is accessible to the Hive connector through the class path and is able to communicate with your custom key management system. If the class implements the org.apache.hadoop.conf.Configurable interface from the Hadoop Java API, the Hadoop configuration is supplied after the object instance is created and before it is invoked to provision or retrieve encryption keys.