ETSI NFV-SOL CNF Auto Scaling With Prometheus via PM Interfaces¶
This document describes how to auto scale CNF in Tacker with Prometheus via Performance Management Interfaces.
Overview¶
The diagram below shows an overview of the CNF auto scaling.
Create PM job
The NFVO sends a request to the Tacker to create a PM job.
Set PM job
Prometheus Plugin sets PM job to Prometheus.
Trigger event
Prometheus collects metrics and decides whether triggering event is needed or not.
POST event
Prometheus sends POST request to Tacker with specified URI. Tacker collects data related to the PM event.
Convert event to report
Tacker receives informed event, converts it to report, and saves it to DB. Tacker also saves timestamp of the event.
Send report notification
VnfPmDriverV2 finds all jobs in the DB and matches the report to job. If there is a job that can match successfully, the report is sent to the specified path of the NFVO. If the match is not successful, the processing ends.
Get PM report
The NFVO make a request for the content of the report, then make a decision of scaling.
Scale
Upon receiving a request to scale VNF from NFVO, tacker-server redirects it to tacker-conductor.
Call Kubernetes API
In tacker-conductor, the request is redirected again to an appropriate infra-driver (in this case Kubernetes infra-driver) according to the contents of the instantiate parameters. Then, Kubernetes infra-driver calls Kubernetes APIs.
Change the number of Pods
Kubernetes Master change the number of Pods according to the API calls.
Prerequisites¶
The following packages should be installed:
tacker
python-tackerclient
At least one VNF instance with status of
INSTANTIATEDis required. You can refer to ETSI NFV-SOL CNF (Containerized VNF) Deployment for the procedure to instantiate VNF.The VNF Package used can refer to the sample.
The following third-party services should be installed
NFVO
Prometheus(including Alertmanager)
Each operator has its own NFVO, there is no restriction here, as long as it conforms to ETSI NFV-SOL 002 v3.3.1 and ETSI NFV-SOL 003 v3.3.1, it can be used.
For the installation of Prometheus and Alertmanager, please refer to the official website.
How to configure Prometheus Plugin¶
The Prometheus Plugin is disabled by default in Tacker.
For it to work, we need to find performance_management in
tacker.conf and change its value to True.
$ vi /etc/tacker/tacker.conf
...
[prometheus_plugin]
performance_management = True
...
After modifying the configuration file, don’t forget to restart the Tacker service to take effect.
$ sudo systemctl stop devstack@tacker
$ sudo systemctl restart devstack@tacker-conductor
$ sudo systemctl start devstack@tacker
How to create a PM job¶
After having a CNF that can scale, we need to create a PM job. It determines the monitoring metrics and monitoring resources to be used by Prometheus.
Note
When having an NFVO client, the request is sent by NFVO.
The interface for creating PM jobs is defined in both ETSI NFV-SOL 002 v3.3.1 and ETSI NFV-SOL 003 v3.3.1.
The following are the parameters required by this interface.
Attribute name |
Data type |
Cardinality |
Description |
|---|---|---|---|
objectType |
String |
1 |
Type of the measured object. The applicable measured object type for a measurement is defined in clause 7.2 of ETSI GS NFV-IFA 027. |
objectInstanceIds |
Identifier |
1..N |
Identifiers of the measured object instances for which performance information is requested to be collected. |
subObjectInstanceIds |
IdentifierInVnf |
0..N |
Identifiers of the measured object instances in case of a structured measured object. |
criteria |
PmJobCriteria |
1 |
Criteria of the collection of performance information. |
>performanceMetric |
String |
0..N |
This defines the types of performance metrics for the specified object instances. Valid values are specified as “Measurement Name” values in clause 7.2 of ETSI GS NFV-IFA 027. At least one of the two attributes (performance metric or group) shall be present. |
>performanceMetricGroup |
String |
0..N |
Group of performance metrics. A metric group is a pre-defined list of metrics, known to the API producer that it can decompose to individual metrics. Valid values are specified as “Measurement Group” values in clause 7.2 of ETSI GS NFV-IFA 027. At least one of the two attributes (performance metric or group) shall be present. |
>collectionPeriod |
UnsignedInt |
1 |
Specifies the periodicity at which the API producer will collect performance information. The unit shall be seconds. |
>reportingPeriod |
UnsignedInt |
1 |
Specifies the periodicity at which the API producer will report to the API consumer. about performance information. The unit shall be seconds. The reportingPeriod should be equal to or a multiple of the collectionPeriod. |
>reportingBoundary |
DateTime |
0..1 |
Identifies a time boundary after which the reporting will stop. The boundary shall allow a single reporting as well as periodic reporting up to the boundary. |
callbackUri |
Uri |
1 |
The URI of the endpoint to send the notification to. |
authentication |
SubscriptionAuthentication |
0..1 |
Authentication parameters to configure the use of Authorization when sending notifications corresponding to this subscription. See as clause 8.3.4 of ETSI GS NFV-SOL 013. |
metadata |
Structure |
1 |
Additional parameters to create PM job. |
>monitoring |
Structure |
1 |
Treats to specify such as monitoring system and driver information. |
>>monitorName |
String |
1 |
In case specifying “prometheus”, backend of monitoring feature is to be Prometheus. |
>>driverType |
String |
1 |
“external”: SCP/SFTP for config file transfer. |
>>targetsInfo |
Structure |
1..N |
Information about the target monitoring system. |
>>>prometheusHost |
String |
1 |
FQDN or ip address of target PrometheusServer. |
>>>prometheusHostPort |
Int |
1 |
Port of the ssh target PrometheusServer. |
>>>alertRuleConfigPath |
String |
1 |
Path of alertRuleConfig path for target Prometheus. |
>>>prometheusReloadApiEndpoint |
String |
1 |
Endpoint url of reload API of target Prometheus. |
>>>authInfo |
Structure |
1 |
Define authentication information to access host. |
>>>>ssh_username |
String |
1 |
The username of the target host for ssh. |
>>>>ssh_password |
String |
1 |
The password of the target host for ssh. |
Note
If
subObjectInstanceIdsis present, the cardinality of theobjectInstanceIdsattribute shall be 1.performanceMetricandperformanceMetricGroup, at least one of the two attributes shall be present.objectTypehas only the following values:Vnf,Vnfc,VnfIntCp,VnfExtCp.
Create PM job can be executed by the following CLI command.
$ openstack vnfpm job create sample_param_file.json --os-tacker-api-version 2
The content of the sample sample_param_file.json in this document is
as follows:
{
"objectType": "Vnf",
"objectInstanceIds": ["495ffedf-2755-42c8-bf14-a5433701311e"],
"subObjectInstanceIds": [],
"criteria": {
"performanceMetric": [
"VcpuUsageMeanVnf.495ffedf-2755-42c8-bf14-a5433701311e"
],
"performanceMetricGroup": [
"VirtualisedComputeResource"
],
"collectionPeriod": 5,
"reportingPeriod": 10,
"reportingBoundary": "2022-06-23T04:56:00.910Z"
},
"callbackUri": "http://localhost:9990/notification/callback/test_performancemanagement_interface_min_1",
"authentication": {
"authType": [
"BASIC",
"OAUTH2_CLIENT_CREDENTIALS"
],
"paramsBasic": {
"userName": "nfvo",
"password": "nfvopwd"
},
"paramsOauth2ClientCredentials": {
"clientId": "auth_user_name",
"clientPassword": "auth_password",
"tokenEndpoint": "token_endpoint"
}
},
"metadata": {
"monitoring": {
"monitorName": "prometheus",
"driverType": "external",
"targetsInfo": [
{
"prometheusHost": "prometheusHost",
"prometheusHostPort": "22",
"authInfo": {
"ssh_username": "ssh_username",
"ssh_password": "ssh_password"
},
"alertRuleConfigPath": "/etc/prometheus/rules/tacker-rule.yml",
"prometheusReloadApiEndpoint": "http://localhost:9990/-/reload"
}
]
}
}
}
Here is an example of create PM job:
$ openstack vnfpm job create sample_param_file.json --os-tacker-api-version 2
+-------------------------+------------------------------------------------------------------------------------------------------------------------+
| Field | Value |
+-------------------------+------------------------------------------------------------------------------------------------------------------------+
| Callback Uri | http://localhost:9990/notification/callback/test_performancemanagement_interface_min_1 |
| Criteria | { |
| | "performanceMetric": [ |
| | "VCpuUsageMeanVnf.495ffedf-2755-42c8-bf14-a5433701311e" |
| | ], |
| | "collectionPeriod": 5, |
| | "reportingPeriod": 10 |
| | } |
| ID | ca9b58cf-8493-44e3-9e76-678ea0e80a80 |
| Links | { |
| | "self": { |
| | "href": "http://127.0.0.1:9890/vnfpm/v2/pm_jobs/ca9b58cf-8493-44e3-9e76-678ea0e80a80" |
| | }, |
| | "objects": [ |
| | { |
| | "href": "http://127.0.0.1:9890/vnflcm/v2/vnf_instances/495ffedf-2755-42c8-bf14-a5433701311e" |
| | } |
| | ] |
| | } |
| Object Instance Ids | [ |
| | "495ffedf-2755-42c8-bf14-a5433701311e" |
| | ] |
| Object Type | Vnf |
| Reports | [] |
| Sub Object Instance Ids | |
+-------------------------+------------------------------------------------------------------------------------------------------------------------+
When creating a PM job, Tacker will modify the configuration file on the
specified Prometheus based on metadata.
Then Prometheus will monitor the specified resource and send the monitored
information to Tacker.
Tacker converts the received monitoring information into a report and
sends a notification request to NFVO.
The following is the request body of a sample notification request.
{
'id': 'c045dae8-cd94-4a29-b09c-96729060f2ad',
'notificationType': 'PerformanceInformationAvailableNotification',
'timeStamp': '2022-09-05T06:56:39Z',
'pmJobId': '34f7a186-88fa-4a42-a35f-30ea9ad710f1',
'objectType': 'Vnf',
'objectInstanceId': 'e30f5f45-522c-4e84-9b2d-9e1669708fff',
'_links': {
'objectInstance': {
'href': 'http://127.0.0.1:9890/vnflcm/v2/vnf_instances/e30f5f45-522c-4e84-9b2d-9e1669708fff'
},
'pmJob': {
'href': 'http://127.0.0.1:9890/vnfpm/v2/pm_jobs/34f7a186-88fa-4a42-a35f-30ea9ad710f1'
},
'performanceReport': {
'href': 'http://127.0.0.1:9890/vnfpm/v2/pm_jobs/34f7a186-88fa-4a42-a35f-30ea9ad710f1/reports/46e95584-7f11-4fd0-b59c-4688c37177ff'
}
}
}
Note
The target URL of this notification request is the Callback Uri
field in the PM job.
How does NFVO Auto Scale CNF¶
NFVO will send a get PM report request to Tacker according to the URL of the report in the notification request.
The response returned by Tacker is as follows:
{
'entries': [
{
'objectType': 'Vnf',
'objectInstanceId': 'e30f5f45-522c-4e84-9b2d-9e1669708fff',
'subObjectInstanceId': 'VDU2-curry-probe-test001-766bdd79bf-wgc7m',
'performanceMetric': 'VCpuUsageMeanVnf.e30f5f45-522c-4e84-9b2d-9e1669708fff',
'performanceValues': [
{
'timeStamp': '2022-09-05T08:02:58Z',
'value': 99.0
}
]
}
]
}
NFVO will determine whether a scale operation is required based on the report data. If needed, a scale request will be sent to Tacker.
How to use the CLI of PM interfaces¶
Create a PM job¶
The creation of PM job has been introduced in the How to create a PM job above, and the use case of the CLI command can be referred to there.
Get all PM jobs¶
Get all PM jobs can be executed by the following CLI command.
$ openstack vnfpm job list --os-tacker-api-version 2
Here is an example of getting all PM jobs:
$ openstack vnfpm job list --os-tacker-api-version 2
+--------------------------------------+-------------+----------------------------------------------------------------------------------------------------------+
| Id | Object Type | Links |
+--------------------------------------+-------------+----------------------------------------------------------------------------------------------------------+
| b8b4095a-148d-42b5-b714-9d703e6c7c62 | Vnf | { |
| | | "self": { |
| | | "href": "http://127.0.0.1:9890/vnfpm/v2/pm_jobs/b8b4095a-148d-42b5-b714-9d703e6c7c62" |
| | | }, |
| | | "objects": [ |
| | | { |
| | | "href": "http://127.0.0.1:9890/vnflcm/v2/vnf_instances/97efce79-34c4-474d-87a0-38ad954f64af" |
| | | } |
| | | ] |
| | | } |
| ca9b58cf-8493-44e3-9e76-678ea0e80a80 | Vnf | { |
| | | "self": { |
| | | "href": "http://127.0.0.1:9890/vnfpm/v2/pm_jobs/ca9b58cf-8493-44e3-9e76-678ea0e80a80" |
| | | }, |
| | | "objects": [ |
| | | { |
| | | "href": "http://127.0.0.1:9890/vnflcm/v2/vnf_instances/495ffedf-2755-42c8-bf14-a5433701311e" |
| | | } |
| | | ] |
| | | } |
+--------------------------------------+-------------+----------------------------------------------------------------------------------------------------------+
Get the specified PM job¶
Get the specified PM job can be executed by the following CLI command.
$ openstack vnfpm job show JOB_ID --os-tacker-api-version 2
Here is an example of getting the specified PM job:
$ openstack vnfpm job show ca9b58cf-8493-44e3-9e76-678ea0e80a80 --os-tacker-api-version 2
+-------------------------+------------------------------------------------------------------------------------------------------------------------+
| Field | Value |
+-------------------------+------------------------------------------------------------------------------------------------------------------------+
| Callback Uri | http://localhost:9990/notification/callback/test_performancemanagement_interface_min_1 |
| Criteria | { |
| | "performanceMetric": [ |
| | "VCpuUsageMeanVnf.495ffedf-2755-42c8-bf14-a5433701311e" |
| | ], |
| | "collectionPeriod": 5, |
| | "reportingPeriod": 10 |
| | } |
| ID | ca9b58cf-8493-44e3-9e76-678ea0e80a80 |
| Links | { |
| | "self": { |
| | "href": "http://127.0.0.1:9890/vnfpm/v2/pm_jobs/ca9b58cf-8493-44e3-9e76-678ea0e80a80" |
| | }, |
| | "objects": [ |
| | { |
| | "href": "http://127.0.0.1:9890/vnflcm/v2/vnf_instances/495ffedf-2755-42c8-bf14-a5433701311e" |
| | } |
| | ] |
| | } |
| Object Instance Ids | [ |
| | "495ffedf-2755-42c8-bf14-a5433701311e" |
| | ] |
| Object Type | Vnf |
| Reports | [ |
| | { |
| | "href": "/vnfpm/v2/pm_jobs/ca9b58cf-8493-44e3-9e76-678ea0e80a80/reports/53aafe25-7124-4880-8b58-47a93b3dc371", |
| | "readyTime": "2022-08-30T08:02:58Z" |
| | } |
| | ] |
| Sub Object Instance Ids | |
+-------------------------+------------------------------------------------------------------------------------------------------------------------+
Change target PM job¶
Updating a PM job can only change two fields, callbackUri and authentication. It can be executed by the following CLI command.
$ openstack vnfpm job update JOB_ID sample_param_file.json --os-tacker-api-version 2
The content of the sample sample_param_file.json in this document is
as follows:
{
"callbackUri": "http://localhost:9990/notification/callback",
"authentication": {
"authType": [
"BASIC",
"OAUTH2_CLIENT_CREDENTIALS"
],
"paramsBasic": {
"userName": "nfvo",
"password": "nfvopwd"
},
"paramsOauth2ClientCredentials": {
"clientId": "auth_user_name",
"clientPassword": "auth_password",
"tokenEndpoint": "token_endpoint"
}
}
}
Here is an example of changing target PM job:
$ openstack vnfpm job update ca9b58cf-8493-44e3-9e76-678ea0e80a80 sample_param_file.json --os-tacker-api-version 2
+-------------------------+------------------------------------------------------------------------------------------------------------------------+
| Field | Value |
+-------------------------+------------------------------------------------------------------------------------------------------------------------+
| Callback Uri | http://localhost:9990/notification/callback |
| Criteria | { |
| | "performanceMetric": [ |
| | "VCpuUsageMeanVnf.495ffedf-2755-42c8-bf14-a5433701311e" |
| | ], |
| | "collectionPeriod": 5, |
| | "reportingPeriod": 10 |
| | } |
| ID | ca9b58cf-8493-44e3-9e76-678ea0e80a80 |
| Links | { |
| | "self": { |
| | "href": "http://127.0.0.1:9890/vnfpm/v2/pm_jobs/ca9b58cf-8493-44e3-9e76-678ea0e80a80" |
| | }, |
| | "objects": [ |
| | { |
| | "href": "http://127.0.0.1:9890/vnflcm/v2/vnf_instances/495ffedf-2755-42c8-bf14-a5433701311e" |
| | } |
| | ] |
| | } |
| Object Instance Ids | [ |
| | "495ffedf-2755-42c8-bf14-a5433701311e" |
| | ] |
| Object Type | Vnf |
| Reports | [] |
| Sub Object Instance Ids | |
+-------------------------+------------------------------------------------------------------------------------------------------------------------+
Delete the specified PM job¶
Delete the specified PM job can be executed by the following CLI command.
$ openstack vnfpm job delete JOB_ID --os-tacker-api-version 2
Here is an example of deleting the specified PM job:
$ openstack vnfpm job delete ca9b58cf-8493-44e3-9e76-678ea0e80a80 --os-tacker-api-version 2
VNF PM job 'ca9b58cf-8493-44e3-9e76-678ea0e80a80' deleted successfully
Get the specified PM report¶
Get the specified PM report can be executed by the following CLI command.
$ openstack vnfpm report show JOB_ID REPORT_ID --os-tacker-api-version 2
Here is an example of getting the specified PM report:
$ openstack vnfpm report show ca9b58cf-8493-44e3-9e76-678ea0e80a80 53aafe25-7124-4880-8b58-47a93b3dc371 --os-tacker-api-version 2
+---------+---------------------------------------------------------------------------------------+
| Field | Value |
+---------+---------------------------------------------------------------------------------------+
| Entries | [ |
| | { |
| | "objectType": "Vnf", |
| | "objectInstanceId": "495ffedf-2755-42c8-bf14-a5433701311e", |
| | "performanceMetric": "VCpuUsageMeanVnf.495ffedf-2755-42c8-bf14-a5433701311e", |
| | "performanceValues": [ |
| | { |
| | "timeStamp": "2022-08-30T08:02:58Z", |
| | "value": "99.0" |
| | } |
| | ] |
| | } |
| | ] |
+---------+---------------------------------------------------------------------------------------+
