Updated on 2025-07-30 GMT+08:00

Creating a GaussDB(DWS) Storage-Compute Decoupled Cluster

storage-compute decoupled uses the cloud-native, cost-effective architecture with decoupled storage and computing. It supports hot and cold data analysis, elastic scaling of storage and computing, unlimited computing power and capacity, and pay-per-use pricing. It is applicable to OLAP analysis scenarios.

This section describes how to create a Decoupled storage and compute cluster on the GaussDB(DWS) console.

Constraints and Limitations

  • You are advised not to use clusters with low specifications, such as clusters with 16 GB memory and 4-core vCPUs, in the production environment. Otherwise, resource overload may occur.
  • To ensure load balancing and high availability for the cluster and prevent service interruptions, you are not advised to connect services directly to a single CN. Instead, configure load balancing to balance the connection to each CN.
  • Storage and compute decoupled clusters do not support database encryption.

Prerequisites

  • You have evaluated the flavor of cluster nodes.

    You can select the number of nodes by data volume, service load, and performance. More nodes bring you stronger storage and compute capabilities.

    When first using GaussDB(DWS), you can create a cluster with a smaller flavor. Then, you can adjust the cluster scale and node flavor based on the data volume and service load changes without interrupting services. For details, see Scaling Out a Cluster.

  • Determine the number of nodes that can be used by users.

    Ensure that the number of available nodes is greater than or equal to 3. Otherwise, the system displays a message indicating that the cluster cannot be created.

    If you have yearly/monthly nodes that meet service requirements, you are advised to use these nodes first to save costs. You can select Yearly/Monthly for Billing Mode.

    If the number of requested nodes, vCPU (cores), or memory (GB) exceed the user's remaining quota, a warning dialog box is displayed, indicating that the quota is insufficient and displaying the detailed remaining quota and the current quota application. You can click Increase quota in the warning dialog box to submit a service ticket and apply for higher node quota. Once approved, we will update your resource quota accordingly and send you a notification. For details, see Quotas.

Procedure

  1. Go to the page for creating a GaussDB(DWS) cluster. Configure the GaussDB(DWS) cluster information as prompted.
  2. Configure the parameters in the Basic Configuration and Data Warehouse Configuration areas for the cluster.
  3. Click Next: Configure Network and configure the parameters in the Network and Others areas.
  4. Click Next: Configure Advanced Settings and configure the parameters in the Cluster Management and Advanced Settings areas.
  5. Click Next: Confirm.
  6. Select a billing mode. If you select the yearly/monthly mode, you also need to configure the service duration. For details, see Yearly/Monthly Subscription.
  7. Click Buy Now. If the billing mode is yearly/monthly billing, click Buy Now. The payment page is displayed.
  8. After the submission is successful, the creation starts. Click Back to Cluster List. The cluster management page is displayed. The initial status of the cluster is Creating. Cluster creation takes some time. Wait for a while. Clusters in the Available state are ready for use.
For how to set the parameters in this area, see Table 1.
Table 1 Parameters in the Basic Configuration area

Parameter

Description

Example Value

Region

Select the AZ for the cluster nodes to run.

For more information about regions, visit Regions and Endpoints.

-

Billing Mode

Select a billing mode as required. For details, see Billing Description.

  • The yearly/monthly billing mode allows you to pay for services upfront before using them, ideal for users who prefer not to make advance payments or long-term commitments.
  • In pay-per-use billing, you only pay for what you use, right down to the minute. This is perfect for users who want to avoid upfront payments and long-term commitments.

Yearly/Monthly

AZ

Select an AZ as required.

For more information, see Regions and AZs.

Single-AZ, AZ 1

For how to set the parameters in this area, see Table 2.
Table 2 Parameters for configuring a data warehouse

Parameter

Description

Example Value

Version

The options are as follows:

Decoupled Storage and Compute: The storage-compute decoupled data warehouse is designed with a cloud native architecture that separates storage and compute. It also features hierarchical auto scaling for computing and storage, as well as multi-logical cluster shared storage technology (Virtual Warehouse or VW). These capabilities allow for computing isolation and concurrent expansion to handle varying loads, making it an ideal choice for OLAP analysis scenarios.

-

CPU Architecture

The following CPU architectures can be selected:

  • x86
  • Kunpeng
NOTE:

The x86 and Kunpeng architectures differ only in their underlying structure, which is not sensible to the application layer. Both architectures use the same SQL syntax. If you need to create a cluster and find that x86 servers are not enough, you can opt for the Kunpeng architecture.

-

Node Flavor

Select a node flavor. Each node flavor shows the vCPU, memory, and recommended application scenario.

For more information about the node flavors supported by GaussDB(DWS) and their prices, see the GaussDB(DWS) pricing details.

For details about the node flavors supported by GaussDB(DWS), see Data Warehouse Specifications.

-

Hot Storage (with Cache)

Available storage capacity of each node.

NOTE:
  • The storage capacity you apply for has the necessary file system overhead, which includes index nodes and the space required for database running. If the idle space is less than 100 GB, scale-out may fail due to insufficient running space.
  • The displayed 200GB/node includes the storage for cache. For example, if you create 3 nodes, each having 200 GB capacity, the total resource capacity is 600 GB, and the actual storage space available to you is 300 GB.

-

Cold Data

Store data in separate OBS buckets, which are billed on a pay-per-use basis.

-

Nodes

Specify the number of nodes in the cluster.

The number of nodes ranges from 3 to 256.

3

Total

Display the cluster's total capacity.

The storage capacity of each flavor includes the storage for cache. The displayed storage capacity includes the disk space consumed by backups and RAIDs.

-

For how to set the parameters in this area, see Table 3.
Table 3 Parameters in the Network area

Parameter

Description

Precaution

Example Value

VPC

Specify a VPC to isolate the cluster's network.

  • If you create a data warehouse cluster for the first time and have not configured the VPC, click View VPC. On the VPC management console that is displayed, create a VPC as needed.

    For how to create a VPC, see Creating a VPC in the Virtual Private Cloud User Guide.

  • After selecting a VPC from the drop-down list, click View VPC to enter the VPC management console and view the detailed information about the VPC.
  • You can click to refresh the options in the VPC drop-down list.

vpc-dws

Subnet

VPC subnet.

A subnet provides dedicated network resources that are isolated from other networks, improving network security.

Once a cluster is created, its subnet cannot be changed. To update the subnet, restore the cluster's snapshot into a new cluster. The new cluster will have identical data, and you can adjust the subnet during its creation.

subnet-dws

Security Group

VPC security group.

A security group restricts access rules to enhance security when GaussDB(DWS) and other services access each other.

  • Automatic creation

    If Automatic creation is selected, the system automatically creates a default security group. This option is selected by default.

    The rule of the default security group is as follows: The outbound allows all access requests, while the inbound is open only to the database port that you set to connect to the GaussDB(DWS) cluster.

    The format of the default security group's name is dws-<cluster-name>-<database-port-of-the-GaussDB(DWS)-cluster>, for example, dws-dws-demo-8000.

  • Manual creation

    You can also log in to the VPC management console to manually create a security group. Then, go back to the page for creating data warehouse clusters, click next to the Security Group drop-down list to refresh the page, and select the new security group.

    To enable the GaussDB(DWS) client to connect to the cluster, you need to add an inbound rule to the new security group to grant the access permission to the database port of the GaussDB(DWS) cluster. The following is an example of an inbound rule. For details, see Adding an Inbound Rule.

    • Protocol: TCP
    • Port: 8000. Use the database port set when the GaussDB(DWS) cluster is created. This port is used by GaussDB(DWS) to receive client connections.
    • Source: Select IP address and use the host IP address of the client host, for example, 192.168.0.10/32.
  • If the quotas of the security group and the security group rule are insufficient, an error message will be displayed after you submit the cluster creation application. You can select an existing group and retry.
  • After a GaussDB(DWS) cluster is created, you can change the security group. You can also add, delete, or modify security group rules in the current security group. For details, see Modifying a Security Group. Changing the security group of a cluster may cause brief service disruption. Exercise caution when performing this operation. To improve network performance, limit your selection to no more than five security groups.

Automatic creation

For how to set the parameters in this area, see Table 4.

Table 4 Parameters in the Others area

Parameter

Description

Precaution

Example Value

EIP

Whether users can use a client to connect to a cluster's database over the Internet. The following methods are supported:

  • Do not use: Do not specify any EIPs here. If GaussDB(DWS) is used in the production environment, first bind it to ELB, and then bind it to an EIP on the ELB page.
  • Buy now: Specify the EIP bandwidth, and an EIP with dedicated bandwidth will be bound to the cluster. The EIP can be used to access the cluster over the Internet. The bandwidth name of an automatically assigned EIP starts with the cluster name.
  • Specify: Specify an EIP to be bound to the cluster. If no available EIPs are displayed in the drop-down list, click View EIP to go to the EIP page and create one that meets your needs. You can set the IP address type and bandwidth as required.
  • In yearly/monthly billing mode, you cannot buy an EIP during cluster creation.
  • The system asks you to create an agency called DWSAccessVPC when you link an EIP to a project in a new region. This allows GaussDB(DWS) to access the VPC. After the authorization is successful, GaussDB(DWS) can switch to a healthy VM when the VM bound with the EIP becomes faulty.
  • By default, only Huawei Cloud accounts or users with Security Administrator permissions can query and create agencies. By default, the IAM users in those accounts cannot query or create agencies. When the users use the EIP, the system makes the binding function unavailable. Contact a user with the DWS Administrator permissions to authorize the agency on the current page. For details, see Allowing GaussDB(DWS) to Manage Resources.
  • Do not use indicates disabling access to the cluster over the public network. After a cluster is created, if you want to access it over the public network, bind an EIP to the cluster and create a public network domain name. For details, see Creating a Public Network Domain Name.
  • If GaussDB(DWS) is used for the production environment, the new GaussDB(DWS) cluster needs to be bound to ELB and then to EIP. Select Do not use here.

Buy Now

Bandwidth

When EIP is set to Buy now, you need to specify the bandwidth of the EIP. The value ranges from 1 Mbit/s to 100 Mbit/s.

-

50Mbit/s

ELB

Whether ELB is bound. With ELB health checks, CN requests of a cluster can be quickly forwarded to normal CNs. If a CN is faulty, the workload can be immediately shifted to a healthy node, minimizing cluster access faults. Currently, ELBs can be bound in the same VPC or across VPCs.

  • Do not use: The load balancer is not used. If GaussDB(DWS) is used in the production environment, first bind it to ELB, and then bind it to an EIP on the ELB page.
  • Specify: Specify an ELB to be bound to the cluster. If no available ELBs are displayed in the drop-down list, click Create ELB to go to the ELB page and create one as needed.

Configure load balancing to ensure load balancing and high availability of the cluster and prevent service interruptions. You are not advised to directly connect services to a single CN.

Specify

For how to set the parameters in this area, see Table 5.

Table 5 Parameters in the Cluster Management area

Parameter

Description

Precaution

Example Value

Cluster Name

Name of the data warehouse cluster.

Enter 4 to 64 characters. Only letters (case-insensitive), digits, hyphens (-), and underscores (_) are allowed. The name must start with a letter.

dws-demo

Cluster Version

Version of the database instance installed in the cluster. The example version number is for reference only.

-

-

Default Database

The default database name of the cluster is gaussdb.

This name cannot be changed.

gaussdb

Administrator Account

Database administrator name.

The administrator username must:

  • Consist of lowercase letters, digits, or underscores.
  • Start with a lowercase letter or an underscore.
  • Contain 6 to 64 characters.
  • The username cannot be a keyword of the GaussDB(DWS) database. For details about the keywords of the GaussDB(DWS) database, see Keyword in the Data Warehouse Service (DWS) Developer Guide.

dbadmin

Administrator Password

Password of the database administrator account.

Change the password regularly and keep it secure.

The password must:
  • Contain 12 to 32 characters.
  • Not be the same as the username or it spelled backwards.
  • Contain at least three of the following character types: uppercase letters, lowercase letters, digits, and special characters (~!?,.:;_(){}[]/<>@#%^&*+|\=-)
  • Pass the weak password check.

-

Confirm Password

Enter the database administrator password again.

-

-

Database Port

Port used when the client or application connects to the database in the cluster.

The port number ranges from 8000 to 30000.

The database port of a created cluster cannot be changed. You can specify the database port only when creating a cluster.

8000

Time Zone

You can set the time zone for the tenant cluster, including the system OS time zone and cluster data warehouse time zone.

-

-

  1. Select the enterprise project of the cluster. You can configure this parameter only when the Enterprise Project Management service is enabled. The default value is default.

    An enterprise project facilitates project-level management and grouping of cloud resources and users.

    You can select the default enterprise project default or other existing enterprise projects. To create an enterprise project, log in to the Enterprise Management console. For details, see the Enterprise Management User Guide.

  2. Configure advanced parameters. Select Default to keep the default values of the advanced parameters. You can also select Custom to modify the values.
    • Backup Device
      Set the backup device used by the current cluster. For details about the parameter configuration principles, see Table 6.
      Table 6 Automated snapshot parameters

      Parameter

      Description

      Backup Device

      Select OBS or NFS from the drop-down list.

      NFS Backup File System Address (NFS)

      NFS shared IP address. To mount the SFS shared path, enter its IP address. If successful, a mount directory will be created in the /var/chroot/nfsbackup directory of the cluster instance.

    • CNs

      CNs, or Coordinators, receive access requests from the clients and return the execution results. They also split and distribute tasks to the data nodes (DNs) for parallel execution.

      The value ranges from 3 to the number of cluster nodes. The maximum value is 20 and the default value is 3. In a large-scale cluster, you are advised to deploy multiple CNs.

    • Tags

      A tag is a key-value pair used to identify a cluster. For details about the keys and values, see Table 7. By default, no tag is added to the cluster.

      If your organization has configured GaussDB(DWS) tag policies, you need to add tags to clusters based on the tag policies. If a tag does not comply with the tag policies, cluster creation may fail. Contact the administrator to learn more about tag policies.

      For details about tags, see Overview.

      Table 7 Tag parameters

      Parameter

      Description

      Key

      A key must be unique in a given cluster. You can:

      • Select a predefined tag key or an existing resource tag key from the drop-down list of the text box. To add a predefined tag, create a predefined tag on TMS and select it from the Key drop-down list. You can click View predefined tags to enter the Predefined Tag page of TMS. Then, click Create Tag to create a predefined tag. For more information, see Creating Predefined Tags in the Tag Management Service User Guide.
      • Enter a tag key in the text box. A tag key can contain a maximum of 128 characters. It cannot be an empty string, start with _sys, or start or end with a space.

        The value can contain only letters, numbers, spaces, and the following characters: _ . : = + - @

      Value

      You can:

      • Select a predefined tag value or resource tag value from the drop-down list of the text box.
      • Enter a tag value in the text box. A tag value can contain a maximum of 255 characters, which can be an empty string. It cannot start or end with a space.

        The value can contain only letters, numbers, spaces, and the following characters: _ . : = + - @

  3. Specify whether to enable the IPv6 dual stack for the cluster. If this function is enabled, a client or application can connect to the database using an IPv6 address.

    To enable IPv6, the following conditions must be met:

    • The subnet configured in Table 3 is an IPv6 dual-stack subnet.
    • The cluster supports IPv6 addresses and should have at least three NICs.
    • The cluster version must be 8.2.1.210 or later.
Table 8 Duration

Parameter

Description

Required Duration (Yearly/Monthly)

Configure the required duration. You get a greater discount if you purchase a longer period. Price is displayed at the bottom of the page for your reference. You can click Pricing details to view the detailed price.

Auto-renewal (Yearly/Monthly)

  • By default, this option is not selected.
  • Renewal rules:
    • Monthly subscriptions are renewed for a month each time.
    • Yearly subscriptions are renewed for a year each time.

    Example: Customer A purchases a cluster in yearly/monthly mode and select enables auto-renewal. If the cluster is subscribed to for eight months, it will be automatically renewed each month. If the cluster is subscribed to for two years, it will be automatically renewed each year. For details about the renewal fee deduction, see Fee Deduction Rules.