Backup Storages

TiDB supports storing backup data to Amazon S3, Google Cloud Storage (GCS), Azure Blob Storage, and NFS. Specifically, you can specify the URL of backup storage in the --storage or -s parameter of br commands. This document introduces the URL format and authentication of different external storage services, and server-side encryption.

Send credentials to TiKV

CLI parameter	Description	Default value
`--send-credentials-to-tikv`	Controls whether to send credentials obtained by BR to TiKV.	`true`

By default, BR sends a credential to each TiKV node when using Amazon S3, GCS, or Azure Blob Storage as the storage system. This behavior simplifies the configuration and is controlled by the parameter --send-credentials-to-tikv(or -c in short).

Note that this operation is not applicable to cloud environments. If you use IAM Role authorization, each node has its own role and permissions. In this case, you need to configure --send-credentials-to-tikv=false (or -c=0 in short) to disable sending credentials:

./br backup full -c=0 -u pd-service:2379 --storage 's3://bucket-name/prefix'

If you back up or restore data using the BACKUP and RESTORE statements, you can add the SEND_CREDENTIALS_TO_TIKV = FALSE option:

BACKUP DATABASE * TO 's3://bucket-name/prefix' SEND_CREDENTIALS_TO_TIKV = FALSE;

URL format

URL format description

This section describes the URL format of the storage services:

[scheme]://[host]/[path]?[parameters]

Amazon S3
GCS
Azure Blob Storage

scheme: s3
host: bucket name
parameters:
- access-key: Specifies the access key.
- secret-access-key: Specifies the secret access key.
- use-accelerate-endpoint: Specifies whether to use the accelerate endpoint on Amazon S3 (defaults to false).
- endpoint: Specifies the URL of custom endpoint for S3-compatible services (for example, <https://s3.example.com/>).
- force-path-style: Use path style access rather than virtual hosted style access (defaults to true).
- storage-class: Specifies the storage class of the uploaded objects (for example, STANDARD or STANDARD_IA).
- sse: Specifies the server-side encryption algorithm used to encrypt the uploaded objects (value options: `, AES256, or aws:kms`).
- sse-kms-key-id: Specifies the KMS ID if sse is set to aws:kms.
- acl: Specifies the canned ACL of the uploaded objects (for example, private or authenticated-read).
- role-arn: When you need to access Amazon S3 data from a third party using a specified IAM role, you can specify the corresponding Amazon Resource Name (ARN) of the IAM role with the role-arn URL query parameter, such as arn:aws:iam::888888888888:role/my-role. For more information about using an IAM role to access Amazon S3 data from a third party, see AWS documentation.
- external-id: When you access Amazon S3 data from a third party, you might need to specify a correct external ID to assume the IAM role. In this case, you can use this external-id URL query parameter to specify the external ID and make sure that you can assume the IAM role. An external ID is an arbitrary string provided by the third party together with the IAM role ARN to access the Amazon S3 data. Providing an external ID is optional when assuming an IAM role, which means if the third party does not require an external ID for the IAM role, you can assume the IAM role and access the corresponding Amazon S3 data without providing this parameter.

scheme: gcs or gs
host: bucket name
parameters:
- credentials-file: Specifies the path to the credentials JSON file on the migration tool node.
- storage-class: Specifies the storage class of the uploaded objects (for example, STANDARD or COLDLINE)
- predefined-acl: Specifies the predefined ACL of the uploaded objects (for example, private or project-private)

scheme: azure or azblob
host: container name
parameters:
- account-name: Specifies the account name of the storage.
- account-key: Specifies the access key.
- access-tier: Specifies the access tier of the uploaded objects, for example, Hot, Cool, or Archive. The value is Hot by default.

URL examples

This section provides some URL examples by using external as the host parameter (bucket name or container name in the preceding sections).

Amazon S3
GCS
Azure Blob Storage

Back up snapshot data to Amazon S3

./br backup full -u "${PD_IP}:2379" \
--storage "s3://external/backup-20220915?access-key=${access-key}&secret-access-key=${secret-access-key}"

Restore snapshot data from Amazon S3

./br restore full -u "${PD_IP}:2379" \
--storage "s3://external/backup-20220915?access-key=${access-key}&secret-access-key=${secret-access-key}"

Back up snapshot data to GCS

./br backup full --pd "${PD_IP}:2379" \
--storage "gcs://external/backup-20220915?credentials-file=${credentials-file-path}"

Restore snapshot data from GCS

./br restore full --pd "${PD_IP}:2379" \
--storage "gcs://external/backup-20220915?credentials-file=${credentials-file-path}"

Back up snapshot data to Azure Blob Storage

./br backup full -u "${PD_IP}:2379" \
--storage "azure://external/backup-20220915?account-name=${account-name}&account-key=${account-key}"

Restore the test database from snapshot backup data in Azure Blob Storage

./br restore db --db test -u "${PD_IP}:2379" \
--storage "azure://external/backup-20220915account-name=${account-name}&account-key=${account-key}"

Authentication

When storing backup data in a cloud storage system, you need to configure authentication parameters depending on the specific cloud service provider. This section describes the authentication methods used by Amazon S3, GCS, and Azure Blob Storage, and how to configure the accounts used to access the corresponding storage service.

Amazon S3
GCS
Azure Blob Storage

Before backup, configure the following privileges to access the backup directory on S3.

Minimum privileges for TiKV and Backup & Restore (BR) to access the backup directories during backup: s3:ListBucket, s3:GetObject, s3:PutObject, s3:DeleteObject, and s3:AbortMultipartUpload
Minimum privileges for TiKV and BR to access the backup directories during snapshot restore: s3:ListBucket and s3:GetObject
Minimum privileges for TiKV and BR to access the backup directories during log restore: s3:ListBucket, s3:GetObject, and s3:PutObject. When restoring log backup data, BR writes the database and table to be restored to the backup directory.

If you have not yet created a backup directory, refer to Create a bucket to create an S3 bucket in the specified region. If necessary, you can also create a folder in the bucket by referring to Create a folder.

It is recommended that you configure access to S3 using either of the following ways:

Method 1: Specify the access key
If you specify an access key and a secret access key in the URL, authentication is performed using the specified access key and secret access key. Besides specifying the key in the URL, the following methods are also supported:
- BR reads the environment variables $AWS_ACCESS_KEY_ID and $AWS_SECRET_ACCESS_KEY.
- BR reads the environment variables $AWS_ACCESS_KEY and $AWS_SECRET_KEY.
- BR reads the shared credentials file in the path specified by the environment variable $AWS_SHARED_CREDENTIALS_FILE.
- BR reads the shared credentials file in the ~/.aws/credentials path.
Method 2: Access based on the IAM role
Associate an IAM role that can access S3 with EC2 instances where the TiKV and BR nodes run. After the association, BR can directly access the backup directories in S3 without additional settings.
```
br backup full --pd "${PD_IP}:2379" \
--storage "s3://${host}/${path}"
```

You can configure the account used to access GCS by specifying the access key. If you specify the credentials-file parameter, the authentication is performed using the specified credentials-file. Besides specifying the key in the URL, the following methods are also supported:

BR reads the file in the path specified by the environment variable $GOOGLE_APPLICATION_CREDENTIALS
BR reads the file ~/.config/gcloud/application_default_credentials.json.
BR obtains the credentials from the metadata server when the cluster is running in GCE or GAE.

Method 1: Specify the access key
If you specify account-name and account-key in the URL, the authentication is performed using the specified access key and secret access key. Besides the method of specifying the key in the URL, BR can also read the key from the environment variable $AZURE_STORAGE_KEY.
Method 2: Use Azure AD for backup and restore
Configure the environment variables $AZURE_CLIENT_ID, $AZURE_TENANT_ID, and $AZURE_CLIENT_SECRET on the node where BR is running.
- When the cluster is started using TiUP, TiKV uses the systemd service. The following example shows how to configure the preceding three environment variables for TiKV:
  Note
  If this method is used, you need to restart TiKV in step 3. If your cluster cannot be restarted, use Method 1: Specify the access key for backup and restore.
  1. Suppose that the TiKV port on this node is 24000, that is, the name of the systemd service is tikv-24000:
```
systemctl edit tikv-24000
```
  2. Edit the TiKV configuration file to configure the three environment variables:
```
[Service]
Environment="AZURE_CLIENT_ID=aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"
Environment="AZURE_TENANT_ID=aaaaaaaa-aaaa-aaaa-aaaa-aaaaaaaaaaaa"
Environment="AZURE_CLIENT_SECRET=aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa"
```
  3. Reload the configuration and restart TiKV:
```
systemctl daemon-reload
systemctl restart tikv-24000
```
- To configure the Azure AD information for TiKV and BR started with command lines, you only need to check whether the environment variables $AZURE_CLIENT_ID, $AZURE_TENANT_ID, and $AZURE_CLIENT_SECRET are configured in the operating environment by running the following commands:
```
echo $AZURE_CLIENT_ID
echo $AZURE_TENANT_ID
echo $AZURE_CLIENT_SECRET
```
- Use BR to back up data to Azure Blob Storage:
```
./br backup full -u "${PD_IP}:2379" \
--storage "azure://external/backup-20220915?account-name=${account-name}"
```

Server-side encryption

Amazon S3 server-side encryption

BR supports server-side encryption when backing up data to Amazon S3. You can also use an AWS KMS key you create for S3 server-side encryption using BR. For details, see BR S3 server-side encryption.

Other features supported by the storage service

BR v6.3.0 supports AWS S3 Object Lock. You can enable this feature to prevent backup data from being tampered with or deleted.