Azure CycleCloud: A Powerful Tool to Orchestrate High-Performance Computing

Azure CycleCloud is a powerful tool designed to help you manage and orchestrate High-Performance Computing (HPC) environments on Azure. Here are some key features and benefits:

1. Scheduler Support: Azure CycleCloud supports various HPC schedulers like Slurm, PBSPro, LSF, Grid Engine, and HTCondor.
2. Dynamic Scaling: It allows for the automatic scaling of clusters based on job load, availability, and time requirements.
3. Resource Management: You can manage virtual machines and scale sets to meet dynamic workload requirements.
4. Integration: CycleCloud integrates with Azure services such as Azure Monitor and Azure Cost Management tools.
5. Ease of Use: It abstracts away the basic Azure building blocks, allowing HPC administrators to focus on managing HPC clusters.

How to create a Slrum cluster using Azure CycleCloud

Creating a Slurm cluster using Azure CycleCloud involves several steps. Here’s a high-level overview to get you started:

1. Install Azure CycleCloud:
   • Ensure you have Azure CycleCloud installed and configured. You can follow the installation guide for detailed instructions.
2. Create a New Cluster:
   • Open the Azure CycleCloud portal.
   • Click on “Create Cluster” and select the Slurm template.
3. Configure the Cluster:
   • Define the cluster name, region, and other basic settings.
   • Configure the scheduler node (master) and execute nodes. For example:
     
[cluster custom-slurm]
[[node scheduler]]
ImageName = cycle.image.centos7
MachineType = Standard_A4
[[[cluster-init cyclecloud/slurm:default]]][[[cluster-init cyclecloud/slurm:scheduler]]][[[configuration]]]
run_list = role[slurm_scheduler_role]
[[nodearray execute]]
ImageName = cycle.image.centos7
MachineType = Standard_A1
[[[cluster-init cyclecloud/slurm:default]]]
[[[cluster-init cyclecloud/slurm:execute]]]
[[[configuration]]]

run_list = role[slurm_scheduler_role]
slurm.autoscale = true
slurm.hpc = true
slurm.default_partition = true

   • Launch the Cluster:
     o After configuring the cluster, click “Create” to launch it. Azure CycleCloud will handle the provisioning and configuration of the nodes.
   • Manage and Monitor:
     o Use the Azure CycleCloud portal to monitor the cluster’s performance, scale nodes up or down, and manage jobs.
   • Run Jobs:
     o Submit jobs to the Slurm scheduler using standard Slurm commands like sbatch, srun, and squeue.

Pricing details

Azure CycleCloud Slurm Workspace pricing is based on the resources you use, such as virtual machines (VMs), storage, and networking. Here are some key points:

1. Virtual Machines (VMs): You pay for the VMs you use, which can vary based on the size and type of VM. Azure offers a range of VM sizes to fit different workloads.
2. Storage: Costs depend on the type and amount of storage you use. Options include Azure NetApp Files and Azure Managed Lustre Filesystem.
3. Networking: Charges may apply for data transfer and network resources like virtual networks and public IP addresses.

Azure CycleCloud itself does not have a separate cost; you only pay for the underlying Azure resources you consume.

Conclusion

Azure CycleCloud is a robust solution for managing and orchestrating High-Performance Computing (HPC) environments on Azure. It simplifies the creation, scaling, and management of HPC clusters, supporting various schedulers like Slurm, PBSPro, and HTCondor. Overall, Azure CycleCloud empowers organizations to efficiently run HPC workloads in the cloud, optimizing performance and cost.

About CloudThat