ESC can auto-scale VMs using the KPI instructions. The scaling workflow begins when the VNF instance is in the instantiated state.The NFVO enables and disables the auto-scaling while modifying isAutoscaleEnabled configurable property of the VNF.

Following are the events that trigger an ETSI-compliant auto-scale, which requires an instigation of a ScaleVnfToLevelRequest: functionality.

• Overload and Underload

  If the state of a VM changes and it is under or overloaded, ESC gets a notification to determine if the scaling is automatically enabled. If it is not, ESC generates a notification towards the ETSI-VNFM component to check the VNF's state.

  The following example shows underloaded notification from ESC:

  Headers:
    esc-status-code = 200
    esc-status-message = VM [sample-dep_vm1_0_fbc3da46-e0c6-40dc-91c8-70b1a88857de] underloaded.
  Body:
  <?xml version="1.0" encoding="UTF-8"?>
  <esc_event xmlns="urn:ietf:params:xml:ns:netconf:base:1.0">
    <deployment_name>sample-dep</deployment_name>
    <event_name>MY_VM_UNDERLOADED</event_name>
    <event_type>VM_UNDERLOADED</event_type>
    <external_deployment_id>e911eecf-5f3f-456c-9c80-d99aca2416da</external_deployment_id>
    <external_tenant_id>etsi_tenant</external_tenant_id>
    <internal_deployment_id>99f7629f-98d3-40f5-ad68-7addcfe07006</internal_deployment_id>
    <internal_tenant_id>etsi_tenant</internal_tenant_id>
    <vm_source>
      <generated_vm_name>sample-dep_vm1_0_fbc3da46-e0c6-40dc-91c8-70b1a88857de</generated_vm_name>
      <interfaces>
        <addresses>
          <address>
            <address_id>0</address_id>
            <gateway>172.24.0.1</gateway>
            <ip_address>172.24.0.37</ip_address>
            <dhcp_enabled>true</dhcp_enabled>
            <prefix>20</prefix>
            <subnet>365a0884-fdb3-424c-afe9-2deb3b39baae</subnet>
          </address>
        </addresses>
        <network_uuid>c7fafeca-aa53-4349-9b60-1f4b92605420</network_uuid>
        <mac_address>fa:16:3e:38:1d:6c</mac_address>
        <nic_id>0</nic_id>
        <port_forwarding/>
        <port_uuid>0aeb9585-5190-4f3b-b1aa-495e09c56b7d</port_uuid>
        <security_groups/>
        <subnet_uuid>none</subnet_uuid>
        <type>virtual</type>
        <vim_interface_name>sample-dep_vm1_0_fbc3da46-e0c6-40dc-91c8-70b1a88857de</vim_interface_name>
      </interfaces>
      <vim_id>default_openstack_vim</vim_id>
      <vim_project>admin</vim_project>
      <vim_project_id>c12f013306d849e5b1bbf257c54d5891</vim_project_id>
      <host_uuid>6b8cf361c5ff08a5a886e26f591b8087dadcf2d2b34fb3b5d2772a8d</host_uuid>
      <host_name>my-server-65</host_name>
      <vm_uuid>9fea3fe7-9417-4734-b962-b24340941ef3</vm_uuid>
      <vm_group_name>vm1</vm_group_name>
      <vm_name>sample-dep_vm1_0_fbc3da46-e0c6-40dc-91c8-70b1a88857de</vm_name>
    </vm_source>
  </esc_event>

  The following example shows overloaded notification from ESC:

  Headers:
    esc-status-code = 200
    esc-status-message = VM [sample-dep_vm1_0_fbc3da46-e0c6-40dc-91c8-70b1a88857de] overloaded.
  Body:
  <?xml version="1.0" encoding="UTF-8"?>
  <esc_event xmlns="urn:ietf:params:xml:ns:netconf:base:1.0">
    <deployment_name>sample-dep</deployment_name>
    <event_name>MY_VM_OVERLOADED</event_name>
    <event_type>VM_OVERLOADED</event_type>
    <external_deployment_id>e911eecf-5f3f-456c-9c80-d99aca2416da</external_deployment_id>
    <external_tenant_id>etsi_tenant</external_tenant_id>
    <internal_deployment_id>99f7629f-98d3-40f5-ad68-7addcfe07006</internal_deployment_id>
    <internal_tenant_id>etsi_tenant</internal_tenant_id>
    <vm_source>
      <generated_vm_name>sample-dep_vm1_0_fbc3da46-e0c6-40dc-91c8-70b1a88857de</generated_vm_name>
      <interfaces>
        <addresses>
          <address>
            <address_id>0</address_id>
            <gateway>172.24.0.1</gateway>
            <ip_address>172.24.0.37</ip_address>
            <dhcp_enabled>true</dhcp_enabled>
            <prefix>20</prefix>
            <subnet>365a0884-fdb3-424c-afe9-2deb3b39baae</subnet>
          </address>
        </addresses>
        <network_uuid>c7fafeca-aa53-4349-9b60-1f4b92605420</network_uuid>
        <mac_address>fa:16:3e:38:1d:6c</mac_address>
        <nic_id>0</nic_id>
        <port_forwarding/>
        <port_uuid>0aeb9585-5190-4f3b-b1aa-495e09c56b7d</port_uuid>
        <security_groups/>
        <subnet_uuid>none</subnet_uuid>
        <type>virtual</type>
        <vim_interface_name>sample-dep_vm1_0_fbc3da46-e0c6-40dc-91c8-70b1a88857de</vim_interface_name>
      </interfaces>
      <vim_id>default_openstack_vim</vim_id>
      <vim_project>admin</vim_project>
      <vim_project_id>c12f013306d849e5b1bbf257c54d5891</vim_project_id>
      <host_uuid>6b8cf361c5ff08a5a886e26f591b8087dadcf2d2b34fb3b5d2772a8d</host_uuid>
      <host_name>my-server-65</host_name>
      <vm_uuid>9fea3fe7-9417-4734-b962-b24340941ef3</vm_uuid>
      <vm_group_name>vm1</vm_group_name>
      <vm_name>sample-dep_vm1_0_fbc3da46-e0c6-40dc-91c8-70b1a88857de</vm_name>
    </vm_source>
  </esc_event>

• VNFD

  The VNFD notification contains the instructions for the scale action required for isAutoscaleEnabled configurable property of the VNF operation flow.

  If the scaling is not enabled automatically, you can instigate the manual LCM operations using the KPI instructions. It is instigated by processing the ESC notification stream. You must validate the notification once you receive the KPI events.

  You must take the following actions:

  • Find the matching VNF instance

  • Validate that the appropriate configuration property is set to enable the automated operation

  If the validation passes then you can request to instigate the operation flow to generate the appropriate operation occurrence and associated notifications. For scaling, any specified KPI data determines the scaling parameters. The properties file includes the following new attributes:

   external.scaling.decision = 1
     #external.scaling.window = 120
     external.healing.decision = 1
     #external.healing.window = 120

• VnfInstance resource

  The VNFD determines the scale level using the current scaleStatus. The processing of the request determines the number of VMs to request from ESCManager. The request only supplies a relative number of increments (SCALE_IN or SCALE_OUT).

  You can call the ScaleVnfToLevel endpoint with the following payload, using vnfInstanceId from the vnfInstance resource of the VNF to be scaled.

  Ensure that the VnfLcmOpOcc.isAutomaticInvocation is set to true.

  The following eexample shows JSON payload:

  {
    /* "instantiationLevelId":"id111", */
    "scaleInfo": [
      { "aspectId":"processing", "scaleLevel":"3" },
      { "aspectId":"database",   "scaleLevel":"2" }
    ]
    "additionalParams": {
      "password": "pass1234",
      "username": "admin"
    },
    "action": "scale_to_level"
  }

ESC can auto-heal VMs using the KPI instructions. The NFVO enables and disables the auto-healing while modifying isAutohealEnabled configurable property of the VNF.

The isAutohealEnabled property permits to enable (TRUE)/disable (FALSE) the auto-healing functionality.