All Products
Search

This Product

    Elastic High Performance Computing:Manually scale out a cluster

    Document Center

    Elastic High Performance Computing:Manually scale out a cluster

    Last Updated:Aug 27, 2026

    If your cluster has insufficient compute nodes for its workload, you can add more.

    Prerequisites

    • The recommended management node configuration and job limits depend on the cluster size. For more information, see Cluster overview.

      Cluster size

      Recommended configuration for management nodes

      Job recommendations

      Number of compute nodes ≤ 100

      • vCPUs ≥ 16

      • Memory ≥ 64 GiB

      • Number of queued jobs < 5,000

      • Number of unfinished jobs < 10,000

      Number of compute nodes ≤ 500

      • vCPUs ≥ 32

      • Memory ≥ 128 GiB

      • Number of queued jobs < 10,000

      • Number of unfinished jobs < 20,000

      Number of compute nodes > 500

      • vCPUs ≥ 64

      • Memory ≥ 256 GiB

      • Number of queued jobs < 10,000

      • Number of unfinished jobs < 20,000

    • You must have an available vSwitch in the target region. For information about how to create a vSwitch, see Create a vSwitch.

    • You must have a sufficient ECS instance quota in the target region. For more information, see Manage ECS quotas.

    Procedure

    1. Log on to the E-HPC console.

    2. In the upper-left corner of the top navigation bar, select a region.

    3. In the left-side navigation pane, click Cluster.

    4. On the Cluster page, find the cluster that you want to scale out and click Expand Capacity.

    5. In the Resize panel, configure the nodes.

      You can scale out a cluster by adding new nodes, existing nodes, or on-premises nodes. Select the appropriate tab and configure the parameters.

      • Add new nodes

        Configure the parameters for the new nodes as described in the following table, and then click Buy Now.

        Parameter

        Description

        Zone

        The new nodes can be in a different zone from the existing nodes in the cluster.

        vSwitch ID

        Select the vSwitch for the new nodes.

        Billing method

        Select a billing method for the new nodes. The following options are available:

        • Subscription: A prepaid billing method. You can purchase and renew resources by week, month, or year.

        • Pay-as-you-go: A postpaid billing method. You are charged hourly based on the actual usage duration.

        • Spot instance: A postpaid billing method. These are on-demand instances available at a lower price than pay-as-you-go instances.

        For more information, see Instance overview.

        Number and instance type of new nodes

        Select the number and instance type for the new compute nodes.

        You can add up to 500 compute nodes at a time. If you need to add more, submit a ticket.

        Image type

        Supported image types include public image, custom image, shared image, Marketplace image, and community image. The available image types depend on the selected region and zone, and the images accessible by your account.

        Image

        Select an image to deploy on the nodes based on the selected image type. Note the following:

        • The image's operating system must be the same as that of the other nodes in the cluster. For example, if the existing nodes in the cluster use a CentOS operating system, you can select only a CentOS image.

        • The major version of the selected image must match that of the other nodes in the cluster. For example, if the cluster uses CentOS 7.x, you must select a CentOS 7.x image.

        • If you select a custom image, ensure it was created from a compute node. If you use a custom image created from a management node or a logon node, errors may occur during the scale-out process.

        Assign public IP address

        If the new nodes need to access the internet, enable this option, and then select a bandwidth billing method and specify the maximum outbound bandwidth.

        Queue

        Select the queue that you want the new nodes to join.

        Hostname prefix, hostname suffix

        Set a prefix and suffix for the node hostnames. When you add multiple nodes at once, setting a hostname prefix and suffix helps you identify and manage the nodes.

        System disk

        Select the type and size of the system disk for the new nodes.

        Data disk

        To attach additional data disks to the new nodes, click Add Data Disk and configure the disk type, size, and quantity.

      • Add existing nodes

        In the instance list, select one or more existing nodes and click OK.

        Note

        Existing nodes must be stopped before they can be added. If you cannot find the node that you want to add, go to the ECS console to check its status.

      • On-premises nodes

        Select a queue, enter the information for each on-premises node individually or import the information in batches, and then click OK.

        Note

        On-premises nodes cannot be moved between queues. If you do not use the default queue, create a dedicated queue for your on-premises nodes.

        The following node information is required:

        • Hostname: Must match the actual hostname of the node.

        • Node IP: Must match the actual IP address of the node.

        • CPU: You can run the lscpu command to view the number of CPUs on the node.

        • Memory: The total memory size of the node.

    Results

    After you scale out the cluster, the system automatically installs and initializes the necessary software on the new compute nodes. This process does not affect existing compute nodes. You can view the status of the added nodes. Choose Resource Management > Nodes, select the cluster that you scaled out, and then select Compute Node from the Node Type drop-down list. A status of Running indicates the node has been successfully added to the cluster.