During the deployment of a VNF, metrics must be defined to instruct the ESC monitoring agent component (MONA) how to determine if the VNF is healthy. The definition of metrics is within the Key Performance Indicator (KPI) section of the VNFD and allow MONA to periodically monitor the VNF to check its health and workload, defined on a per-VNFC basis. Actions are then associated with these KPIs and executed when the appropriate conditions are met.

There are several built-in monitoring methods such as ICMP Ping and SNMP. Some of the metrics to monitor on the constituent VNFCs include:

• reachability

• resource usage (such as CPU, memory, disk and network throughput)

The following pre-requisites must be met for the deployed VNFCs to be monitored:

• The deployed VNFCs must be alive

• Monitoring is enabled

• KPIs must be configured

Example:

vdu1:
      type: tosca.nodes.nfv.Vdu.Compute
      properties:
        name: Example VDU1
        description: Example VDU
        boot_order:
          - boot1-volume
        configurable_properties:
          additional_vnfc_configurable_properties:
            vim_flavor: Automation-Cirros-Flavor
            bootup_time: 1800
            vm_name_override: my-vdu-1
            recovery_action: REBOOT_THEN_REDEPLOY
            recovery_wait_time: 100
            monitor_on_error: false
            max_retries: 2
            kpi_data:
              VM_ALIVE-1:
                event_name: 'VM_ALIVE-1'
                  metric_value: 1
                  metric_cond: 'GT'
                  metric_type: 'UINT32'
                  metric_occurrences_true: 1
                  metric_occurrences_false: 30
                  metric_collector: 
                    type: 'ICMPPing'
                    nicid: 1
                    poll_frequency: 10
                    polling_unit: 'seconds'
                    continuous_alarm: false
            admin_rules:
              VM_ALIVE-1:
                event_name: 'VM_ALIVE-1'
                action:
                  - 'ALWAYS log'
                  - 'FALSE recover autohealing'
                  - 'TRUE esc_vm_alive_notification'
            placement_type: zone
            placement_target: nova
            placement_enforcement: strict
            vendor_section:
              cisco_esc:
                config_data:
                  example.txt:
                    file: ../Files/Scripts/example.txt
                      variables:
                        DOMAIN_NAME: { get_input: DOMAIN_NAME }
                        NAME_SERVER: { get_input: NAME_SERVER }
                        VIP_ADDR: { get_input: VIP_ADDR }
                        VIP_PREFIX: { get_input: VIP_PREFIX }
        vdu_profile:
          min_number_of_instances: 1
          max_number_of_instances: 1
               capabilities:
        virtual_compute:
          properties:
            virtual_cpu:
              num_virtual_cpu: 8
            virtual_memory:
              virtual_mem_size: 16
      requirements:
        - virtual_storage:  cdr1-volume
        - virtual_storage:  boot1-volume

The kpi_data shown above is the default KPI required that is required in all deployments at a minimum so that the VM_ALIVE message is generated to tell ESC Manager that the VNFC has been deployed successfully; it consists of the KPI, how it is collected and the actions to be executed when the KPI is met.

Cisco data structure properties

# For notificationType "InfrastructureOperationOccurrenceNotification"
subscription.notifications.infra.filter.operationTypes=MONITORING_MIGRATION
subscription.notifications.infra.filter.operationStates=COMPLETED,FAILED_TEMP,FAILED,ROLLED_BACK
subscription.notifications.infra.callbackUri=http://<nfvoHost>:<nfvoPort>/monitoring/migration/notification # Full URL where the notification will be sent
subscription.notifications.infra.authentication.authType=BASIC # or OAUTH2_CLIENT_CREDENTIALS

# Basic Auth credentials (based on authType)
subscription.notifications.infra.authentication.paramsBasic.userName=nfvo
subscription.notifications.infra.authentication.paramsBasic.password=mypw

# Alternatively, OAUTH 2.0 credentials (based on authType)
#subscription.notifications.infra.authentication.paramsOauth2ClientCredentials.clientId=
#subscription.notifications.infra.authentication.paramsOauth2ClientCredentials.clientPassword=
#subscription.notifications.infra.authentication.paramsOauth2ClientCredentials.tokenEndpoint=

Note	If the previous properties are not set, then these notifications are sent to the subscribers where all notifications types are applicable.

For more information on KPIs and Rules, see the Cisco Elastic Services Controller User Guide.

You can set and unset monitoring of VMs using RESTful interface.

The operation is defaults to asynchronous, you must set sync.supported=true to use this functionaity in a synchronous way.

A payload is required to monitor VMs:

Method type

POST

VNFM Endpoint

Example for SOL003:

{apiRoot}/or_vnfm/vnflcm/v2/ext/vnf_instances/{vnfInstanceId}/monitoring/operations

{apiRoot}/ve_vnfm/vnflcm/v2/ext/vnf_instances/{vnfInstanceId}/monitoring/operations

To start and stop monitoring operation on a specified VM, set the vnfcInstanceIds

with payload:

{
    "vnfcInstanceIds": ["vnfcInstanceId1","vnfcInstanceId2",...,"vnfcInstaceIdN"],    ## optional
    "operation": "ENABLE_MONITOR",                                                    ## mandatory ENABLE_MONITOR, DISABLE_MONITOR, REBOOT
    "additionalParams": []                                                            ## optional - for future use :-)
}

To start and stop monitoring operation on the entire VNF, do not set the vnfcInstanceIds.

You must mention enable_monitoring to set VM monitoring, and disable_monitoring to unset VM monitoring in the operation field.

Note	When a user reboots the VM from the ESC ETSI interface, the monitoring is automatically enabled.