VNF scaling with reserved resources¶
Tacker allows you to configure reserved compute resources in reservation
policy. The compute resources should be first reserved in the OpenStack
Blazar service by creating leases which can then be configured in the
VNFD template.
TOSCA schema for reservation policy¶
Tacker defines TOSCA schema for the reservation policy as given below:
tosca.policies.tacker.Reservation:
derived_from: tosca.policies.Reservation
reservation:
start_actions:
type: map
entry_schema:
type: string
required: true
before_end_actions:
type: map
entry_schema:
type: string
required: true
end_actions:
type: map
entry_schema:
type: string
required: true
properties:
lease_id:
type: string
required: true
Following TOSCA snippet shows VNFD template using reservation policy. In this policy, you can see there are three different types of actions.
- start_actions
- before_end_actions
- end_actions
In these actions, you can configure multiple actions but scaling policy is mandatory in start_actions and one of before_end_actions or end_actions. The scaling policy configured in the start_actions will be scaling-out policy so configure max_instances as per the compute resources reserved in the Blazar service and the scaling policy configured in either of before_end_actions or end_actions will be scaling-in policy so configure min_instances to 0. Also, default_instances should be set to 0 because we don’t want VDUs until tacker receives the lease start trigger from Blazar through Aodh service. The parameter increment should also be set equal to max_instances as tacker will receive lease start trigger only once during the lifecycle of a lease.
policies:
- RSV:
type: tosca.policies.tacker.Reservation
reservation:
start_actions: [SP_RSV, log]
before_end_actions: [SP_RSV]
end_actions: [noop]
properties:
lease_id: { get_input: lease_id }
- SP_RSV:
type: tosca.policies.tacker.Scaling
properties:
increment: 2
cooldown: 120
min_instances: 0
max_instances: 2
default_instances: 0
targets: [VDU1]
Installation and configurations¶
- You need Blazar, ceilometer and Aodh OpenStack services.
- Modify the below configuration files:
/etc/blazar/blazar.conf:
[oslo_messaging_notifications]
driver = messaging, log
/etc/ceilometer/event_pipeline.yaml:
sinks:
- name: event_sink
transformers:
publishers:
- gnocchi://?archive_policy=low&filter_project=gnocchi_swift
- notifier://
- notifier://?topic=alarm.all
/etc/ceilometer/event_definitions.yaml:
- event_type: lease.event.start_lease
traits: &lease_traits
lease_id:
fields: payload.lease_id
project_id:
fields: payload.project_id
user_id:
fields: payload.user_id
start_date:
fields: payload.start_date
end_date:
fields: payload.end_date
- event_type: lease.event.before_end_lease
traits: *lease_traits
- event_type: lease.event.end_lease
traits: *lease_traits
Deploying reservation tosca template with tacker¶
When reservation resource type is virtual:instance¶
- Create a lease in blazar for instance reservation:
$ blazar lease-create --reservation resource_type=virtual:instance,vcpus=1,memory_mb=1024,disk_gb=20,amount=0,affinity=False
--start-date "2019-04-24 20:00" --end-date "2019-07-09 21:00" lease-1
+--------------+-----------------------------------------------------------------+
| Field | Value |
+--------------+-----------------------------------------------------------------+
| created_at | 2018-12-10 07:44:46 |
| degraded | False |
| end_date | 2019-07-09T21:00:00.000000 |
| events | { |
| | "status": "UNDONE", |
| | "lease_id": "aca14613-2bed-480e-aefe-97fa02813fcf", |
| | "event_type": "start_lease", |
| | "created_at": "2018-12-10 07:44:49", |
| | "updated_at": null, |
| | "time": "2019-04-24T20:00:00.000000", |
| | "id": "038c882a-1c9e-4785-aab0-07a6898653cf" |
| | } |
| | { |
| | "status": "UNDONE", |
| | "lease_id": "aca14613-2bed-480e-aefe-97fa02813fcf", |
| | "event_type": "before_end_lease", |
| | "created_at": "2018-12-10 07:44:49", |
| | "updated_at": null, |
| | "time": "2019-07-09T20:00:00.000000", |
| | "id": "607fb807-55e1-44ff-927e-64a4ec71b0f1" |
| | } |
| | { |
| | "status": "UNDONE", |
| | "lease_id": "aca14613-2bed-480e-aefe-97fa02813fcf", |
| | "event_type": "end_lease", |
| | "created_at": "2018-12-10 07:44:49", |
| | "updated_at": null, |
| | "time": "2019-07-09T21:00:00.000000", |
| | "id": "fd6b1f91-bfc8-49d8-94a7-5136ee2fdaee" |
| | } |
| id | aca14613-2bed-480e-aefe-97fa02813fcf |
| name | lease-1 |
| project_id | 683322bea7154651b18792b59df67d4e |
| reservations | { |
| | "status": "pending", |
| | "memory_mb": 1024, |
| | "lease_id": "aca14613-2bed-480e-aefe-97fa02813fcf", |
| | "resource_properties": "", |
| | "disk_gb": 10, |
| | "resource_id": "bb335cc1-770d-4251-90d8-8f9ea95dac56", |
| | "created_at": "2018-12-10 07:44:46", |
| | "updated_at": "2018-12-10 07:44:49", |
| | "missing_resources": false, |
| | "server_group_id": "589b014e-2a68-48b1-87ee-4e9054560206", |
| | "amount": 1, |
| | "affinity": false, |
| | "flavor_id": "edcc0e22-1f7f-4d57-abe4-aeb0775cbd36", |
| | "id": "edcc0e22-1f7f-4d57-abe4-aeb0775cbd36", |
| | "aggregate_id": 6, |
| | "vcpus": 1, |
| | "resource_type": "virtual:instance", |
| | "resources_changed": false |
| | } |
| start_date | 2019-04-24T20:00:00.000000 |
| status | PENDING |
| trust_id | 080f059dabbb4cb0a6398743abcc3224 |
| updated_at | 2018-12-10 07:44:49 |
| user_id | c42317bee82940509427c63410fd058a |
+--------------+-----------------------------------------------------------------+
2. Replace the flavor, lease_id and server_group_id value in the parameter file
given for reservation with the lease response flavor, lease_id and
server_group_id value.
Ref:
samples/tosca-templates/vnfd/tosca-vnfd-instance-reservation-param-values.yaml
Note
The server_group_id parameter should be specified in VDU section only when reservation resource type is virtual:instance. Operator shouldn’t configure both placement policy under policies and server_group_id in VDU in VNFD template otherwise the server_group_id specified in VDU will be superseded by the server group that will be created by heat for placement policy.
{
flavor: 'edcc0e22-1f7f-4d57-abe4-aeb0775cbd36',
lease_id: 'aca14613-2bed-480e-aefe-97fa02813fcf',
resource_type: 'virtual_instance',
server_group_id: '8b01bdf8-a47c-49ea-96f1-3504fccfc9d4',
}
Sample tosca-template:
tosca_definitions_version: tosca_simple_profile_for_nfv_1_0_0
description: VNF TOSCA template with flavor input parameters
metadata:
template_name: sample-tosca-vnfd-instance-reservation
topology_template:
inputs:
flavor:
type: string
description: Flavor Information
lease_id:
type: string
description: lease id
resource_type:
type: string
description: reservation resource type
server_group_id:
type: string
description: server group id
node_templates:
VDU1:
type: tosca.nodes.nfv.VDU.Tacker
properties:
image: cirros-0.4.0-x86_64-disk
flavor: { get_input: flavor }
reservation_metadata:
resource_type: { get_input: resource_type }
id: { get_input: server_group_id }
CP1:
type: tosca.nodes.nfv.CP.Tacker
properties:
management: true
order: 0
anti_spoofing_protection: false
requirements:
- virtualLink:
node: VL1
- virtualBinding:
node: VDU1
VL1:
type: tosca.nodes.nfv.VL
properties:
network_name: net_mgmt
vendor: Tacker
policies:
- RSV:
type: tosca.policies.tacker.Reservation
reservation:
start_actions: [SP_RSV]
before_end_actions: [SP_RSV]
end_actions: [noop]
properties:
lease_id: { get_input: lease_id }
- SP_RSV:
type: tosca.policies.tacker.Scaling
properties:
increment: 2
cooldown: 120
min_instances: 0
max_instances: 2
default_instances: 0
targets: [VDU1]
Scaling process
After the lease lifecycle begins in the Blazar service, tacker will receive a
start_lease event at 2019-04-24T20:00:00. Tacker will start scaling-out
process and you should notice VDUs will be created as per the increment
value.
Similarly, when before_end_lease event is triggered at 2019-07-09T20:00,
tacker will start scaling-in process in which VDUs will be deleted as per the
increment value.
When reservation resource type is physical:host¶
- Create a lease for compute host reservation:
$ blazar lease-create --physical-reservation min=1,max=1,hypervisor_properties='[">=", "$vcpus", "2"]' --start-date
"2019-04-08 12:00" --end-date "2019-07-09 12:00" lease-1
+--------------+--------------------------------------------------------------+
| Field | Value |
+--------------+--------------------------------------------------------------+
| created_at | 2018-12-10 07:42:44 |
| degraded | False |
| end_date | 2019-07-09T12:00:00.000000 |
| events | { |
| | "status": "UNDONE", |
| | "lease_id": "5caba925-b591-48d9-bafb-6b2b1fc1c934", |
| | "event_type": "before_end_lease", |
| | "created_at": "2018-12-10 07:42:46", |
| | "updated_at": null, |
| | "time": "2019-07-09T11:00:00.000000", |
| | "id": "62682a3a-07fa-49f9-8f95-5b1d8ea49a7f" |
| | } |
| | { |
| | "status": "UNDONE", |
| | "lease_id": "5caba925-b591-48d9-bafb-6b2b1fc1c934", |
| | "event_type": "end_lease", |
| | "created_at": "2018-12-10 07:42:46", |
| | "updated_at": null, |
| | "time": "2019-07-09T12:00:00.000000", |
| | "id": "9f98f8a3-3154-4e8f-b27e-8f61646110d2" |
| | } |
| | { |
| | "status": "UNDONE", |
| | "lease_id": "5caba925-b591-48d9-bafb-6b2b1fc1c934", |
| | "event_type": "start_lease", |
| | "created_at": "2018-12-10 07:42:46", |
| | "updated_at": null, |
| | "time": "2019-04-08T12:00:00.000000", |
| | "id": "c9cd4310-ba8e-41da-a6a0-40dc38702fab" |
| | } |
| id | 5caba925-b591-48d9-bafb-6b2b1fc1c934 |
| name | lease-1 |
| project_id | 683322bea7154651b18792b59df67d4e |
| reservations | { |
| | "status": "pending", |
| | "before_end": "default", |
| | "lease_id": "5caba925-b591-48d9-bafb-6b2b1fc1c934", |
| | "resource_id": "1c05b68f-a94a-4c64-8010-745c3d51dcd8", |
| | "max": 1, |
| | "created_at": "2018-12-10 07:42:44", |
| | "min": 1, |
| | "updated_at": "2018-12-10 07:42:46", |
| | "missing_resources": false, |
| | "hypervisor_properties": "[\">=\", \"$vcpus\", \"2\"]", |
| | "resource_properties": "", |
| | "id": "c56778a4-028c-4425-8e99-babc049de9dc", |
| | "resource_type": "physical:host", |
| | "resources_changed": false |
| | } |
| start_date | 2019-04-08T12:00:00.000000 |
| status | PENDING |
| trust_id | dddffafc804c4063898f0a5d2a6d8709 |
| updated_at | 2018-12-10 07:42:46 |
| user_id | c42317bee82940509427c63410fd058a |
+--------------+--------------------------------------------------------------+
2. Replace the flavor with reservation in tosca-tempate given for reservation
policy as below:
Ref:
samples/tosca-templates/vnfd/tosca-vnfd-host-reservation.yaml
Note
reservation id will be used only when reservation resource type is
physical:host.
Add lease_id and reservation id in the parameter file.
{
resource_type: 'physical_host',
reservation_id: 'c56778a4-028c-4425-8e99-babc049de9dc',
lease_id: '5caba925-b591-48d9-bafb-6b2b1fc1c934',
}
Sample tosca-template:
tosca_definitions_version: tosca_simple_profile_for_nfv_1_0_0
description: VNF TOSCA template with reservation_id input parameters
metadata:
template_name: sample-tosca-vnfd-host-reservation
topology_template:
inputs:
resource_type:
type: string
description: reservation resource type
reservation_id:
type: string
description: Reservation Id Information
lease_id:
type: string
description: lease id
node_templates:
VDU1:
type: tosca.nodes.nfv.VDU.Tacker
properties:
image: cirros-0.4.0-x86_64-disk
reservation_metadata:
resource_type: { get_input: resource_type }
id: { get_input: reservation_id }
CP1:
type: tosca.nodes.nfv.CP.Tacker
properties:
management: true
order: 0
anti_spoofing_protection: false
requirements:
- virtualLink:
node: VL1
- virtualBinding:
node: VDU1
VL1:
type: tosca.nodes.nfv.VL
properties:
network_name: net_mgmt
vendor: Tacker
policies:
- RSV:
type: tosca.policies.tacker.Reservation
reservation:
start_actions: [SP_RSV]
before_end_actions: [noop]
end_actions: [SP_RSV]
properties:
lease_id: { get_input: lease_id }
- SP_RSV:
type: tosca.policies.tacker.Scaling
properties:
increment: 2
cooldown: 120
min_instances: 0
max_instances: 2
default_instances: 0
targets: [VDU1]
Scaling process
After the lease lifecycle begins in the Blazar service, tacker will receive a
start_lease event at 2019-04-08T12:00:00. Tacker will start scaling-out
process and you should notice VDUs will be created as per the increment
value.
Similarly, when end_lease event is triggered at 2019-07-09T12:00, tacker
will start scaling-in process in which VDUs will be deleted as per the
increment value.
