Authenticate and Authorize Users
ScalarDB Cluster has a mechanism to authenticate and authorize users.
This guide describes how to use authentication and authorization in ScalarDB Cluster.
Overview
By using authentication and authorization, you can create users and grant or revoke their privileges. You can create a user by using the CREATE USER
command, and you can grant or revoke one's privileges on a table or a namespace by using the GRANT
or REVOKE
command, respectively. For details about such data control language (DCL) commands, see DCL.
Users can log in to ScalarDB Cluster with a username and a password and execute SQL statements if they have the required privileges.
Authentication and authorization support two types of users:
- Superusers: This type of user has all privileges. Only superusers can create or drop other users and namespaces.
- Normal users: This type of user initially doesn't have any privileges, so they need to be granted privileges by a superuser or another user who has the
GRANT
privilege.
The following privileges are available when using authentication and authorization:
SELECT
INSERT
UPDATE
DELETE
CREATE
DROP
TRUNCATE
ALTER
GRANT
For details about privileges, see Which privileges are required for each type of operation.
Configurations
This section describes the available configurations for authentication and authorization.
ScalarDB Cluster node configurations
To enable authentication and authorization, you need to set scalar.db.cluster.auth.enabled
to true
.
Name | Description | Default |
---|---|---|
scalar.db.cluster.auth.enabled | Whether authentication and authorization are enabled. | false |
You can also set the following configurations:
Name | Description | Default |
---|---|---|
scalar.db.cluster.auth.cache_expiration_time_millis | Cache expiration time for authentication and authorization information in milliseconds. | 60000 (1 minute) |
scalar.db.cluster.auth.auth_token_expiration_time_minutes | Authentication and authorization token expiration time in minutes. | 1440 (1 day) |
scalar.db.cluster.auth.auth_token_gc_thread_interval_minutes | Authentication and authorization token garbage collection (GC) thread interval in minutes. | 360 (6 hours) |
scalar.db.cluster.auth.pepper | A secret value added to a password before hashing. If not specified, the password is hashed without pepper. |
If you enable authentication and authorization, you will also need to set scalar.db.cross_partition_scan.enabled
to true
for the system namespace (scalardb
by default) because authentication and authorization perform cross-partition scans internally.
ScalarDB Cluster Java client SDK configurations
To enable authentication and authorization on the client side, you need to set scalar.db.cluster.auth.enabled
to true
.
Name | Description | Default |
---|---|---|
scalar.db.cluster.auth.enabled | Whether authentication and authorization are enabled. | false |
In addition to the configuration in the ScalarDB Cluster SQL client configurations section, you also need to set scalar.db.sql.cluster_mode.username
and scalar.db.sql.cluster_mode.password
to specify the username and password of the client.
Name | Description | Default |
---|---|---|
scalar.db.sql.cluster_mode.username | The username of the client. | |
scalar.db.sql.cluster_mode.password | The password of the client. |
Initial user
When you enable authentication and authorization, the initial user admin
is created and the initial password of that user is admin
. This user is a superuser and has all privileges. You can log in with this user and create other users if necessary.
For security purposes, be sure to change the password of the initial user, especially before deploying to a production environment.
Which privileges are required for each type of operation
The following tables show which privileges are required for each type of operation:
DDL
Command | Superuser required | Required privileges |
---|---|---|
CREATE NAMESPACE | true | |
DROP NAMESPACE | true | |
CREATE TABLE | CREATE | |
DROP TABLE | DROP | |
CREATE INDEX | CREATE | |
DROP INDEX | DROP | |
TRUNCATE TABLE | TRUNCATE | |
ALTER TABLE | ALTER | |
CREATE COORDINATOR TABLES | true | |
DROP COORDINATOR TABLES | true | |
TRUNCATE COORDINATOR TABLES | true |
DML
Command | Superuser required | Required privileges |
---|---|---|
SELECT | SELECT | |
INSERT | INSERT | |
UPSERT | INSERT | |
UPDATE | SELECT and UPDATE | |
DELETE | SELECT and DELETE |
DCL
Command | Superuser required | Required privileges |
---|---|---|
CREATE USER | true | |
ALTER USER | true (Users can change their own password.) | |
DROP USER | true | |
GRANT | GRANT (Users can grant only the privileges that they have.) | |
REVOKE | GRANT (Users can revoke only the privileges that they have.) |
Wire encryption
ScalarDB Cluster also supports wire encryption by using Transport Layer Security (TLS). If you enable authentication and authorization, enabling wire encryption in production environments to protect the user credentials is strongly recommended.
This wire encryption feature encrypts:
- The communications between the ScalarDB Cluster node and clients.
- The communications between all ScalarDB Cluster nodes (the cluster's internal communications).
This feature uses gRPC's TLS support. For details, see the official gRPC Security Policy.
Configurations
This section describes the available configurations for wire encryption.
ScalarDB Cluster node configurations
To enable wire encryption, you need to set scalar.db.cluster.tls.enabled
to true
.
Name | Description | Default |
---|---|---|
scalar.db.cluster.tls.enabled | Whether wire encryption (TLS) is enabled. | false |
You also need to set the following configurations:
Name | Description | Default |
---|---|---|
scalar.db.cluster.tls.ca_root_cert_pem | The custom CA root certificate (PEM data) for TLS communication. | |
scalar.db.cluster.tls.ca_root_cert_path | The custom CA root certificate (file path) for TLS communication. | |
scalar.db.cluster.tls.override_authority | The custom authority for TLS communication. This doesn't change what host is actually connected. This is intended for testing, but may safely be used outside of tests as an alternative to DNS overrides. For example, you can specify the hostname presented in the certificate chain file that you set for scalar.db.cluster.node.tls.cert_chain_path . | |
scalar.db.cluster.node.tls.cert_chain_path | The certificate chain file used for TLS communication. | |
scalar.db.cluster.node.tls.private_key_path | The private key file used for TLS communication. |
To specify the certificate authority (CA) root certificate, you should set either scalar.db.cluster.tls.ca_root_cert_pem
or scalar.db.cluster.tls.ca_root_cert_path
. If you set both, scalar.db.cluster.tls.ca_root_cert_pem
will be used.
ScalarDB Cluster Java client SDK configurations
To enable wire encryption on the client side, you need to set scalar.db.cluster.tls.enabled
to true
.
Name | Description | Default |
---|---|---|
scalar.db.cluster.tls.enabled | Whether wire encryption (TLS) is enabled. | false |
You also need to set the following configurations:
Name | Description | Default |
---|---|---|
scalar.db.cluster.tls.ca_root_cert_pem | The custom CA root certificate (PEM data) for TLS communication. | |
scalar.db.cluster.tls.ca_root_cert_path | The custom CA root certificate (file path) for TLS communication. | |
scalar.db.cluster.tls.override_authority | The custom authority for TLS communication. This doesn't change what host is actually connected. This is intended for testing, but may safely be used outside of tests as an alternative to DNS overrides. For example, you can specify the hostname presented in the certificate chain file that you set for scalar.db.cluster.node.tls.cert_chain_path . |
To specify the CA root certificate, you should set either scalar.db.cluster.tls.ca_root_cert_pem
or scalar.db.cluster.tls.ca_root_cert_path
. If you set both, scalar.db.cluster.tls.ca_root_cert_pem
will be used.