OASIS Topology and Orchestration Specification for Cloud Applications (TOSCA) TC

 View Only
Back to discussions
Expand all | Collapse all

RE: [tosca] Introducing Turandot: A TOSCA orchestrator for Kubernetes

  • 1.  RE: [tosca] Introducing Turandot: A TOSCA orchestrator for Kubernetes

    Posted 04-22-2020 22:16




    Hi Tal,
     
    Thanks for sharing. I ve asked Chris to brief us on his new TOSCA orchestrator at tomorrow s TC meeting, after a walkthrough of the new CSD01 candidate. Would you be interested in doing the same for Turandot
    at our May 12 th TC meeting?
     
    Thanks again,
    Paul
     

    From: tosca@lists.oasis-open.org <tosca@lists.oasis-open.org>
    On Behalf Of Tal Liron
    Sent: Wednesday, April 22, 2020 6:02 PM
    To: tosca@lists.oasis-open.org
    Subject: [tosca] Introducing Turandot: A TOSCA orchestrator for Kubernetes

     


    This one has been a long time coming!


     


    The
    Puccini project is intentionally designed to purely address the issue of TOSCA and not have any opinion about orchestration. The approach there is "BYOO": Bring Your Own Orchestrator.


     


    Well, now it's time to bring one in!


     


    The
    Turandot orchestrator is, at its essence, Puccini running as a
    Kubernetes operator . Using either a built-in inventory, or a robust external inventory product, one can register CSARs as annotated service templates, and then instantiate them into Kubernetes resources.


     


    Included is a TOCA profile for Kubernetes, which very closely matches the Kubernetes specs themselves with some TOSCA quality-of-life details added on top. KubeVirt is given the same treatment, and indeed one can use it to run full-blown
    virtual machines within Kubernetes. Multus is also supported for advanced networking via Kubernetes CNI.


     


    TOSCA policies govern workload composition via TOSCA substitution mapping. Multi-cluster workloads are handled by "delegates" -- Turandot can install itself on a another cluster, send the service templates there, and let the remote Turandot
    operator manage its Kubernetes resources.


     


    Included is a
    complex multi-cluster MANO example , which showcases an abstract network service that uses provisioning policies to substitute the network functions for VNFs, CNFs, and even an out-of-cluster PNF. The data plane is itself a node template that can be substituted
    for various technologies.


     


    As of now, Turandot is not quite ready for everyone to use. Documentation is lacking, and there need to be many more (and simpler) examples. However, it can be demoed. :) If there's interest, I would be more than happy to show it in action
    in an upcoming TOSCA TC meeting.








  • 2.  Re: [tosca] Introducing Turandot: A TOSCA orchestrator for Kubernetes

    Posted 04-22-2020 22:19
    Sure! On Wed, Apr 22, 2020 at 5:16 PM Paul L. < paul.lipton@live.com > wrote: Hi Tal, Thanks for sharing. I ve asked Chris to brief us on his new TOSCA orchestrator at tomorrow s TC meeting, after a walkthrough of the new CSD01 candidate. Would you be interested in doing the same for Turandot at our May 12 th TC meeting? Thanks again, Paul From: tosca@lists.oasis-open.org < tosca@lists.oasis-open.org > On Behalf Of Tal Liron Sent: Wednesday, April 22, 2020 6:02 PM To: tosca@lists.oasis-open.org Subject: [tosca] Introducing Turandot: A TOSCA orchestrator for Kubernetes This one has been a long time coming! The Puccini project is intentionally designed to purely address the issue of TOSCA and not have any opinion about orchestration. The approach there is "BYOO": Bring Your Own Orchestrator. Well, now it's time to bring one in! The Turandot orchestrator is, at its essence, Puccini running as a Kubernetes operator . Using either a built-in inventory, or a robust external inventory product, one can register CSARs as annotated service templates, and then instantiate them into Kubernetes resources. Included is a TOCA profile for Kubernetes, which very closely matches the Kubernetes specs themselves with some TOSCA quality-of-life details added on top. KubeVirt is given the same treatment, and indeed one can use it to run full-blown virtual machines within Kubernetes. Multus is also supported for advanced networking via Kubernetes CNI. TOSCA policies govern workload composition via TOSCA substitution mapping. Multi-cluster workloads are handled by "delegates" -- Turandot can install itself on a another cluster, send the service templates there, and let the remote Turandot operator manage its Kubernetes resources. Included is a complex multi-cluster MANO example , which showcases an abstract network service that uses provisioning policies to substitute the network functions for VNFs, CNFs, and even an out-of-cluster PNF. The data plane is itself a node template that can be substituted for various technologies. As of now, Turandot is not quite ready for everyone to use. Documentation is lacking, and there need to be many more (and simpler) examples. However, it can be demoed. :) If there's interest, I would be more than happy to show it in action in an upcoming TOSCA TC meeting.


Related Content