HPC auto scaling feature to Azure cloud RRS feed

  • Question

  • Dear all,

    In our environment, we are going to have a hybrid HPC installation where a head node and few computational nodes will reside in our premise, and additional computational nodes will be in the Azure cloud. I tried to add a computational node in the cloud and was able to do it manually.

    For us, when cloud-based computational nodes are not running any jobs we will be wasting money. So we want to enable/disable the cloud computational nodes based on the computational needs. Is there any way we can accomplish this with HPC automatically without any manual intervention? Based on the number of jobs we can automatically scale up/down our cloud computational nodes to save the cost. Please let us know if it's possible with HPC.



    Puneet Sharma

    Tuesday, March 28, 2017 9:56 PM

All replies

  • hi

      Auto grow shrink is an area we are keep improving. For what's available in the released version, please check https://docs.microsoft.com/zh-cn/azure/virtual-machines/windows/classic/hpcpack-cluster-node-autogrowshrink

    Qiufang Shi

    Tuesday, March 28, 2017 11:57 PM
  • Dear Qiufang,

    Instead of Azure cloud, if we want to have Windows HPC on google/amazon cloud, can we have this auto grow shrink feature of HPC? Please let us know.



    Puneet Sharma

    Tuesday, April 11, 2017 11:07 AM
  • Hi Pruneet,

       Auto grow shrink in google/amazon cloud is not support out of the box. But it is possible but requires a few work. If you look at the auto grow shrink script metioned in this doc, the script will call three powershell script:

    - Add-HPCIaaSNode.ps1

    - Start-HPCIaaSNode.ps1

    - Stop-HPCIaaSNode.ps1

    - Remove-HPCIaaSNode.ps1

    you need re-implement these powershell script against other cloud provider so that it works for you. And in HPC Pack 2016, we supported compute node without domain join. Thus, it will be easier for you to add/remove/start/stop these type of compute nodes (Including linux nodes) in the cloud.

    Qiufang Shi

    Wednesday, April 12, 2017 2:22 AM