this is the DM agreed to R2 and maintenance release page version 20.

the final R3 DM is still under discussion.


any change to this page done by someone other then  Victor Gao or Michael Lando will be reverted thank you.

Capabilities Type
tosca.capabilites.nfv.VirtualBindable 
tosca.capabilities.nfv.VirtualBindable:
  derived_from: tosca.capabilities.root
tosca.capabilities.nfv.Metric
tosca.capabilites.nfv.VirtualCompute 
tosca.capabilities.nfv.VirtualCompute:
  derived_from: tosca.capabilities.Root
  properties:
    logical_node:
      type: tosca.datatypes.nfv.LogicalNodeData
      required: false
    compute_requirements: 
      type: map
      required: false
    requested_additional_capabilities:
      type: map
      entry_schema:
         type: tosca.datatypes.nfv.RequestedAdditionalCapability
      required: false
    virtual_memory:
      type: tosca.datatypes.nfv.VirtualMemory
      required: true
    virtual_cpu:
      type: tosca.datatypes.nfv.VirtualCpu
      required: true
tosca.capabilities.nfv.VirtualLinkable 
tosca.capabilities.nfv.VirtualLinkable:
  derived_from: tosca.capabilities.root
tosca.capabilities.nfv.ExtVirtualLinkable 
tosca.capabilities.nfv.ExtVirtualLinkable:
  derived_from: tosca.capabilities.root
tosca.capabilities.nfv.VirtualStorage 
tosca.capabilities.nfv.VirtualStorage:
  derived_from: tosca.capabilities.root




  • No labels

18 Comments

  1. shitao li

    We also need to check the defintion of tosca.capabilities.nfv.Metric in OASIS TOSCA, and figour out why it is derived from Endpoint capability.

  2. Andrei Kojukhov

    Regarding the Monitoring feature we also need to consider the ONAP VES use case where the monitoring events should be handled. To my understanding there is no applicable support in IFA011 spec. 

  3. maopeng zhang

    Monitor model here is not enough to be coded and is TBD later.  

    I see some capablities derived from tosca.capabilities.Node and some derived from tosca.capabilities.Root. Is there some principle?

    1. shitao li

      Good question. I think, they should all be derived from tosca.capabilities.Root, not tosca.capabilities.Node.

      1. maopeng zhang

        All listed derived from Root maybe are reasonable. 

        1. Alexander Vul

          We need to align with the specs, not with what is reasonable... (smile)

          1. maopeng zhang

            So I do not change it, and let us to discuss the issues. If it is a problem, maybe we need to submit the issues. Now the SOL is also a draft, not a final one.

          2. shitao li

            we need to change the spec as well. I will do next week.

            1. Arturo Martin de Nicolas

              So you are bring this up to the TOSCA YAML, right? Personally I don't see either what's the purpose on deriving some capabilities from capabilities.node instead of from capabilities.root.

            2. maopeng zhang

              If discuss the issue, please info me. thanks

  4. Lianhao Lu

    I think one thing is missing in tosca.capabilities.nfv.VirtualCompute, the ephemeral root local disk size, which means the amount of disk space to use for the root (/) partition to be launched with for a VM. This information is required to create OpenStack flavors.

    1. maopeng zhang

      My suggestions:

      1, The local_disk as a property in the virtualComputer. such as simple tosca compute node.

      2,  or add a local_disk node, related capacity and relationship,...

      In the local disk, mostly it is created with the computer. Option 1 is OK in my mind.

      1. Haibin Chu

        I also think option 1 is a better approach. Besides that, the name of "local_disk" sounds a little bit confusing to me, because image could also be a local disk for the virtual compute. Maybe we can just call it "ephemeral" to align the name in VIMs, e.g. Openstack.

  5. Alexander Vul

    Someone keeps changing the definitions of the HPA attributes from 

            type: map

            required: xxxx

    to

           type: map

           entry_schema:

               entry: string

           required: xxxx


    Why?


    1. Alexander Vul

      maopeng zhang -  is there a reason you changed the type of vdu_storage requirements as described above?

      1. maopeng zhang

        Alexander Vul  If defined a map in tosca, the entry_schema is needed

  6. shitao li

    HI Alex

    I think maopeng is correct. Based on TOSCA grammar, for defining a key/value pair, the type of the property should be "map" and the entry_schema is string.  


    <entry_key_1>:   <entry_value_1>

    ...

    <entry_key_n>:   <entry_value_n>


    for example:

    <some_entity>:

      ...

      properties: 

        emails:

          type: map

          entry_schema:

            description:   basic email address

            type: string

            constraints:

              -   max_length: 128

  7. Thinh Nguyenphu

    It was brought to my attention that tosca.capabilities.nfv.VirtualStorage was not included.  Thus, I update the wiki page with tosca.capabilities.nfv.VirtualStorage definition.