All Products
Search

This Product

    ApsaraDB RDS:Download backups

    Document Center

    ApsaraDB RDS:Download backups

    Last Updated:Oct 27, 2025

    This topic describes how to download backups from ApsaraDB RDS for MySQL for scenarios such as long-term retention, backup queries, backup migration, and auditing.

    Before you begin

    • No fees are charged for failed tasks.

    • You cannot cancel a backup download task after it starts.

    • You cannot use downloaded backup data to directly restore an RDS instance. To restore an instance, see Overview of data restoration solutions.

    • Directly downloading a backup is not always the best option. Before you start, identify your instance type and scenario to choose the most efficient and cost-effective method.

    Billing details

    You may be charged for outbound network traffic when you download backups from an instance that uses high-performance local disks. For instances that use cloud disks, you may be charged for both outbound network traffic and backup set conversion.

    Note

    If you download backups to OSS, you are charged based on your actual usage of OSS.

    Network traffic fees

    • Internal network downloads: Free of charge.

    • External network downloads: Each instance receives a free quota of 500 GB per month. Usage beyond this quota is billed on a pay-as-you-go basis. You can view the downloaded traffic volume on the Basic Information page of the RDS instance. For more information about unit prices, see Network fees.

    Note

    • You can download backups over an internal network or the internet for instances that use cloud disks and instances that use high-performance local disks.

    • When you start a snapshot backup download for an instance that uses cloud disks, the system creates a data conversion task. You cannot obtain a download link while the task is in progress or has failed. No outbound network traffic fees are incurred during this period. After the task is successfully completed, you are charged for the actual traffic generated when you use the download link.

    • Physical backup downloads for instances that use high-performance local disks do not involve data conversion. You can obtain the download link directly. You are charged for the actual traffic generated when you use the link to download the backup.

    • To view the public network backup download volume, go to the RDS Instance List page, click the instance ID, and then view Backup Download Volume in the Instance Resources section on the Basic Information page.

    • We recommend that you purchase a subscription network plan to cover data transfer fees for downloading backups over the public network. The larger the network plan capacity, the higher the discount.

    Backup set conversion fees

    This fee applies when you use the advanced download feature for an instance that uses cloud disks to convert a snapshot backup into a CSV (without header), CSV with header, SQL, or Parquet file. No free quota is provided. The billing standards are as follows:

    Region

    Conversion fee (USD/GB)

    Public cloud

    0.03125

    Note

    When you start a snapshot backup download for an instance that uses cloud disks, the system creates a data conversion task. No backup set conversion fees are charged if this task fails.

    Cloud disk instance download

    Instances that use cloud disks support advanced download tasks for a specific point in time or a specific backup set. You can convert snapshot backups into CSV, SQL, or Parquet format. You can also download the backup to a URL or save the backup file directly to an OSS bucket.

    Prerequisites

    Before you start, make sure that the instance meets all the following conditions.

    • Instance status: Running.

    • Storage class: ESSD or premium performance disk.

    • Database version: MySQL 8.0 or 5.7.

    • Region support: China (Chengdu), China (Shenzhen), China (Guangzhou), China (Qingdao), China (Beijing), China (Zhangjiakou), China (Hangzhou), China (Shanghai), China (Hong Kong), Malaysia (Kuala Lumpur), Indonesia (Jakarta), Japan (Tokyo), Singapore, US (Silicon Valley), US (Virginia), and Germany (Frankfurt). This feature is being gradually rolled out to other regions. The regions displayed in the console are the ones that are supported.

    • Minor engine version: The minor engine version must be later than 20201031. If the minor engine version of your instance does not meet this requirement, you must upgrade the minor engine version.

      Important

      If the Download Instance Backup button is unavailable even though the minor engine version meets the requirements, this may be because the database proxy is enabled. Some instances do not support this download feature due to hardware limitations. To resolve this issue, you must first disable the database proxy and then upgrade the minor engine version. After the issue is resolved, you can re-enable the proxy service as needed.

    • Backup encryption: Disabled. The advanced download feature is not supported for encrypted instances that use cloud disks.

    • Data archiving: Disabled. The advanced download feature is not supported for instances that use premium performance disks for which data archiving is enabled.

    • RAM permissions: The RAM user must have permissions to download backup files. For more information about how to grant permissions to a RAM user, see Grant permissions to a read-only RAM user to download backup files.

    Feature limitations

    • Unsupported objects: The downloaded backup files do not include expression indexes, foreign keys, generated columns, hidden columns, views, functions, stored procedures, system variables, or triggers.

    • Unsupported databases: The download does not include system databases such as information_schema, mysql, performance_schema, sys, or __recycle_bin__.

    • Unsupported data types: The download task may fail if a table contains special data types such as GEOMETRY, POINT, LINESTRING, POLYGON, MULTIPOINT, MULTILINESTRING, MULTIPOLYGON, or GEOMETRYCOLLECTION.

    • Storage class restriction: If the download destination is OSS, the storage class of the target bucket must be Standard. If you want to save the backup to an OSS bucket that uses a storage class other than Standard, you can use one of the following methods:

      • Method 1: Download the backup to a bucket that uses the Standard storage class, and then transform the storage class.

      • Method 2: Download the backup to a local machine, and then upload it to the destination bucket.

    Procedure

    1. Navigate to the backup list

      1. Go to the RDS instance list, select the region where the instance is located, and then click the instance ID.

      2. In the navigation pane on the left, choose Backup and Restoration > Base Backup List.

    2. Start the download task

      • Download a log backup: On the Log Backup tab, find the target log file and click Download in the Actions column to obtain the download link. The log backup feature must be enabled.

      • Download a data backup: On the Data Backup tab, find the target backup set and click Download Instance Backup in the Actions column.

    3. Configure the parameters for the instance backup download

      1. In the Download Time Point and Backup Set step, select Download by Backup Set or Download by Time Point, and then click Next.

      2. In the Download Instance and Database/Table step, keep the default setting (Instance Download) and click Next.

      3. In the Download Destination and Format step, configure the download destination and format.

        Download Destination

        Configuration

        OSS (Recommended)

        1. Enter the bucket name and directory prefix of the destination OSS.

        2. Set Download Format to CSV (without header), CSV-with-header, SQL, or Parquet.

        3. If this is your first time using this feature, click Go to Authorize > Agree to Authorization as prompted to grant RDS permissions to access OSS.

        URL

        1. Select URL.

        2. Select the download format.

        3. After the task is successful, you must manually generate a download link. The link is valid for a limited time.

    4. Confirm and execute: Read and select the acknowledgment, and then click Finish. The page automatically redirects to the Backup Download List tab, where you can monitor the task status.

    5. Obtain the backup file

      • OSS method: Wait for the task status to change to Successful. The file is automatically downloaded to the specified OSS bucket and directory.

      • URL method: Wait for the task status to change to Successful. Within the validity period of 3 days, click Generate Link in the Download Destination column. Set the link validity period, which can be from 5 minutes to 1 day and is 2 hours by default, to obtain the internal or external download URL. After the task and the link expire, the temporary data is automatically deleted. You must start a new download task if the link expires.

      Important

      The conversion process competes for computing resources on the service side. The download may fail due to temporary resource allocation failures or unsupported data formats. If a task fails, try to download the backup again.

    6. Use a command-line tool to download the file

      After you obtain the URL download link:

      • Use a command-line tool such as wget or curl to download the file. This is the recommended method. For more information, see Appendix 1: Download commands.

      • If the file is small, you can copy the external URL to your browser's address bar.

    Download backups from an instance that uses high-performance local disks

    The process for downloading backups from an instance that uses high-performance local disks is more direct. This process does not require data format conversion, and you can immediately obtain the download link for a physical backup or a log backup.

    Prerequisites

    Before you start, make sure that the instance meets all the following conditions.

    Procedure

    1. Navigate to the backup list

      1. Go to the RDS instance list, select the region where the instance is located, and then click the instance ID.

      2. In the navigation pane on the left, choose Backup and Restoration > Base Backup List.

    2. Obtain the download link

      1. Select the Data Backup or Log Backup tab.

      2. Find the target backup file and click Download Instance Backup (for a data backup) or Download (for a log backup) in the Actions column.

      3. In the dialog box that appears, you can directly copy the internal URL or the external URL.

        • Internal URL: This is the recommended method. You must download the backup from an Alibaba Cloud ECS instance that is in the same region and VPC. This method is fast and free of charge. Cross-region downloads over the internal network are not supported.

        • External URL: Use this to download from a local machine or another network environment.

    3. Use a command-line tool to download the file

      After you obtain the URL download link:

      • Use a command-line tool such as wget or curl to download the file. This is the recommended method. For more information, see Appendix 1: Download commands.

      • If the file is small, you can copy the external URL to your browser's address bar.

    Note

    • The backup download link is valid for 1 hour. Download the file promptly to prevent the link from expiring. If the link expires, you must start a new download task to obtain a new link. A download task that has already started is not affected by the link's validity period.

    • Log backup refers to the backup of local logs. To obtain local logs, see mysqlbinlog.

    • If an instance is a High-availability Edition or a three-node edition, both the primary and secondary instances generate log backups. You can view the IDs of the primary and secondary instances on the Service Availability page.

    • If you want to restore data to a self-managed database, the point in time of the data backup must be within the time range of the log backup. The log backup and data backup must belong to the same instance. This means they must have the same instance ID.

    Appendix 1: Download commands

    After you obtain the URL download link, we recommend that you use a command-line tool such as wget or curl to download the file. These tools support resumable downloads and help ensure stability.

    Note

    • If the network download speed is too low (less than 64 KB/s), the download may be interrupted. Make sure you have a stable network connection during the download.

    • If you download a backup set to an ossfs-mounted disk, you may need to adjust the multipart_size parameter. By default, this parameter supports a maximum file size of 100 GB. If a backup set is larger than 100 GB, the download fails. For more information about ossfs and its configuration, see ossfs and Mounting options.

    • We recommend that you use the wget or curl commands provided in this topic to download backup sets. If you use other third-party tools, the file may be downloaded repeatedly (the actual downloaded data size is larger than the backup set size). This will incur extra fees for the outbound traffic that exceeds the backup set size.

    wget

    nohup wget -c -t 0 "Backup file download URL" -O Path and filename to save the downloaded file > Download process monitoring log &

    Parameters:

    Parameter

    Description

    nohup

    Prevents the download from being interrupted if the terminal disconnects or if you accidentally perform a copy operation. The process automatically exits after the download is complete.

    -t

    Number of retries. Set to 0 for infinite retries.

    -c

    Supports resumable downloads.

    -O

    The path and filename to save the downloaded file.

    Example:

    nohup wget -c -t 0 "https://dbs-****.aliyundoc.com/****.tar.gz****" -O /backup/examplebackup.tar.gz > /tmp/download.log &

    curl

    nohup curl -C - --retry 10 "Backup file download URL" -o Path and filename to save the downloaded file > Download process monitoring log &

    Parameters:

    Parameter

    Description

    nohup

    Prevents the download from being interrupted if the terminal disconnects or if you accidentally perform a copy operation. The process automatically exits after the download is complete.

    --retry

    Number of retries if the task fails. Set to 10 to retry 10 times.

    -C -

    Supports automatic resumable downloads.

    -o

    The path and filename to save the downloaded file.

    Example:

    nohup curl -C - --retry 10 "https://dbs-****.aliyundoc.com/****.tar.gz****" -o examplebackup.tar.gz > /tmp/download.log &

    Appendix 2: Scenarios

    Scenario and purpose

    Recommendation

    • Migrate from an RDS instance to another RDS instance or a self-managed database

    • Restore an existing backup

    Choose one of the following methods:

    • Query data from a specific point in time

    • Query data within a backup

    Choose one of the following methods:

    • Restore the backup to a pay-as-you-go or Serverless instance, query the data, and then release the instance. For more information, see Restore full data.

    • Follow the instructions in this topic to download the backup and restore it to a self-managed database for querying.

    • Save a local copy of a backup

    • Download a backup for auditing

    Choose one of the following methods:

    Upload a backup to OSS

    RDS backups are stored in backup storage and do not occupy instance storage space. To store backups in OSS, you can choose one of the following methods:

    • For eligible instances that use cloud disks, you can create an advanced download task as described in this topic and choose to save the downloaded data directly to OSS during the task configuration. For other instances that use cloud disks or instances that use high-performance local disks, you can first download the backup as described in this topic and then upload it to OSS.

    • Use DBS to create a backup and select User OSS as the storage type. You can then directly use Data Disaster Recovery to restore the backup data from OSS to RDS.

    Retain backups long-term

    Automatically download backups

    Download binary logs

    References

    FAQ

    Can ApsaraDB RDS database backups be uploaded to OSS?

    Yes, they can. ApsaraDB RDS for MySQL instances that use cloud disks support creating advanced download tasks for a point in time or a specific backup set. You can also save the downloaded data directly to your OSS. For more information, see Download backups from an instance that uses cloud disks in this topic. You can also download backups from an ApsaraDB RDS for MySQL instance that uses cloud disks or high-performance local disks to a local machine and then upload large files to OSS.

    Is there a bandwidth limit for downloading MySQL backups over the internet?

    No, there is not. ApsaraDB RDS itself does not have a bandwidth limit. The speed of downloading backups over the internet depends on your server's bandwidth. If your server has limited bandwidth, the backup download speed may be restricted. In addition, when you download backup data using an external link, you must pay for outbound traffic that exceeds the free quota. For more information, see Billing details.

    How do I use the downloaded data backups and log backups?

    You can restore the backup files downloaded from an instance that uses high-performance local disks to a self-managed database. For more information, see Restore an ApsaraDB RDS for MySQL physical backup file to a self-managed database or Restore an ApsaraDB RDS for MySQL logical backup file to a self-managed database.

    Can the default qp.xb backup format for physical backups of instances that use high-performance local disks be converted to .gz format?

    • If the Restore to database or table feature is enabled for the instance, the backup compression format must be qp.xb. The format cannot be converted.

    • If the Restore to database or table feature is not enabled for the instance, you can use the ModifyBackupPolicy API operation to change the backup set compression format from qp.xb to tar.gz. This operation only changes the compression method of the backup set and does not affect the instance in any other way.

    When I restore data downloaded from an instance that uses cloud disks to a local MySQL database, the error ERROR 1148 (42000): The used command is not allowed with this MySQL version is reported. How do I fix this?

    Run the query script show variables like 'local_infile'; on MySQL. If the result is OFF, run the following statement to enable file import: set global local_infile = 1;. After this operation is complete, run the import script again.

    Why is the backup size displayed in the console different from the actual size of the downloaded backup file?

    The downloaded backup file is compressed and is generally smaller than the backup size displayed in the console. You can restore the file and check if the data is complete. For more information, see Restore an ApsaraDB RDS for MySQL snapshot backup file to a self-managed database.

    Can a backup file downloaded from the ApsaraDB RDS console be directly restored to another Alibaba Cloud RDS instance?

    No. A backup file downloaded from the ApsaraDB RDS console cannot be directly restored to a new RDS instance. You can:

    Can a backup download task in progress be canceled?

    No, you cannot. You cannot cancel a backup download task after it starts.

    What should I do if some data is missing from the backup list?

    You can perform the following steps:

    1. Restore the full data to a new instance.

    2. On the new instance, check and confirm which tables or data are missing.

    3. Export the missing tables or data from the original instance.

    4. Use DTS to synchronize the exported data from the original instance back to the restored new instance.

    When I use an internal URL to download a backup, the error "ERROR 403: Forbidden." is reported. How do I fix this?

    1. Check whether the instance meets the conditions for an internal network connection. Make sure that the ECS instance in the VPC can communicate with RDS over the internal network.

    2. Check whether the format of the download command is correct. For example, check whether the download URL uses full-width double quotation marks.

    If you still cannot access RDS over the internal network, try using an external URL to download the backup.

    How do I delete expired backup information from the backup download list?

    Information in the backup download list cannot be deleted. This does not affect the use of the instance.

    Can a downloaded backup file be directly imported into a self-managed database?

    No, it cannot. A physical backup cannot be directly imported into a self-managed database. To import the backup, we recommend that you manually create a logical backup, download it to a local machine, and then import it.

    When I use the wget command to download a backup, the error "ERROR 504: Gateway Time-out." is reported. How do I fix this?

    When you use the wget command, we recommend that you add the --read-timeout=300 --connect-timeout=300 parameter.

    Does downloading a backup use the bandwidth of the current instance?

    No.

    Downloading over an internal network does not affect the instance's bandwidth. Downloading over the internet is only billed and depends only on your local network bandwidth.

    Why is the file garbled after I use a download command to download a backup?

    After the backup is downloaded to a local machine or an ECS instance, run the tar -izxvf <compressed_package_filename>.tar.gz -C <path_to_decompressed_file> command to decompress the file and view it in the corresponding folder.

    How are the downloaded backup files named?

    The backup files are named after the host ID and the time of the instance.

    When I download a backup file to OSS, the error "Unsupported bucket XXX's StorageClass is IA, Please Use bucket's StorageClass is Standard" is reported. How do I fix this?

    When you download a backup to OSS, the storage class of the target bucket must be Standard. If you want to save the downloaded backup to an OSS bucket that uses a different storage class, you can use one of the following methods:

    • Method 1: Download the backup to a bucket that uses Standard storage, and then transform the storage class.

    • Method 2: Download the backup to a local machine, and then upload it to the destination bucket.

    Why can't I find the backup download buttons in the console?

    • If your instance is a primary instance, it must meet the prerequisites for backup download.

    • If your instance is a read-only instance, you must download the backup from its corresponding primary instance because read-only instances are not backed up.

    Why am I charged even though I did not download any backups?

    You were likely charged the Backup set conversion fee. This fee is incurred when you use the advanced download feature for a disk instance to convert a snapshot backup into a CSV file (with or without a header), SQL file, or Parquet file. You can view the Backup Conversion Traffic in the Backup Download List.

    For more information about downloading backups, see FAQ about backup downloads.