Manage ML Clusters

Note

The ML Clusters feature is only available when enable-ml-cluster is set to true in the deployment configuration. For details, see Deploy SynxDB Cloud.

To enable advanced data analytics and AI capabilities within SynxDB Cloud such as training machine learning models, fine-tuning large language models (LLMs), or deploying AI inference services, you need to provision dedicated computing resources. These resources are managed as ML Clusters.

Platform administrators use DBaaS Admin Console to view and manage these clusters, ensuring that data scientists and developers have the necessary infrastructure to run SynxML workloads efficiently.

Access the ML Clusters panel

  1. Log in to the DBaaS Admin Console.

  2. In the left navigation menu, click Organizations.

  3. In the organization tree, expand the desired organization and click on the target account.

  4. The right-hand side displays the Account Detail page.

  5. At the bottom of this page, click the ML Clusters tab.

View existing ML Clusters

You can see all created ML clusters under the account with details in the ML Cluster tab, including:

  • Name

  • Minimum value

  • Maximum value

  • Status

  • Stop/Start toggle

  • Operation buttons (Edit, Delete)

Create an ML Cluster

  1. On the ML Clusters tab, click + Create ML Cluster in the top-right corner.

  2. In the Create ML Cluster dialog, fill in the following fields:

    • Name (Required): Enter a unique name for the ML cluster.

    • Owner (Required): Select the owner of the cluster from the dropdown list.

    • Count (Required): Specify the Minimum value and Maximum value for the number of replicas to allow for auto-scaling.

    • Warehouse (Required): Select the warehouse to associate with this ML cluster.

    • Profile (Required): Select a resource profile that defines the CPU, memory, and disk storage specifications.

    • Environment Spec: Choose an environment specification if applicable.

    • Undelegated Bucket: Toggle this switch to use an external object storage bucket (for example, from a public cloud provider) that has been registered in the DBaaS Admin Console. If disabled, the cluster uses the default internal object storage resources.

  3. Click OK to create the cluster.

Manage existing ML clusters

You can perform the following operations on existing ML clusters in the ML Clusters list:

  • Start/Stop: Toggle the switch in the Stop/Start column to enable or disable the cluster.

  • Edit: Click Edit in the Operation column to modify the cluster’s configuration, such as scaling the replica count or updating the profile.

  • Delete: Click Delete in the Operation column to permanently remove the cluster.

What’s next

After creating an ML cluster, you can use it to run SynxML workloads. See SynxML SQL for more information.