This page describes the component specification (JSON) sections that are common to both Docker and CDAP components. Differences for each are pointed out below. Elements that are very different, and described in the CDAP or Docker specific pages.
The component specification is represented (and validated) against this Component Spec json schema and described below:
The "Meta Schema" implementation defines how component specification JSON schemas can be written to define user input. It is itself a JSON schema (thus it is a "meta schema"). It requires the name of the component entry, component type (either 'cdap' or 'docker') and a description under "self" object. The meta schema version must be specified as the value of the "version" key. Then the input schema itself is described.
The table below shows the four types of schema descriptions supported There are four types of schema descriptions objects - jsonschema for inline standard JSON Schema definitions of JSON inputs, delimitedschema for delimited data input using a JSON description defined by AT&T, unstructured for unstructured text, and reference that allows a pointer to another artifact for a schema. The reference allows for XML schema, but can be used as a pointer to JSON, Delimited Format, and Unstructured schemas as well.
Metadata refers to the properties found under the self
JSON. This group of properties is used to uniquely identify this component specification and identify the component that this specification is used to capture.
Example:
"self": { "version": "1.0.0", "name": "asimov.component.kpi_anomaly", "description": "Classifies VNF KPI data as anomalous", "component_type": "docker" },
self
Schema:
Property Name | Type | Description |
---|---|---|
version | string | Required. Semantic version for this specification |
name | string | Required. Full name of this component which is also used as this component's catalog id. |
description | string | Required. Human-readable text describing the component and the components functional purpose. |
component_type | string | Required. Identify what containerization technology this component uses: docker or cdap . |
Interfaces are the JSON objects found under the streams
key and the services
key. These are used to describe the interfaces that the component uses and the interfaces that the component provides. The description of each interface includes the associated data format.
The streams
JSON is for specifying data produced for consumption by other components, and the streams expected to subscribe to that is produced by other components. These are "fire and forget" type interfaces where the publisher of a stream does not expect or parse a response from the subscriber.
The term stream
here is abstract and neither refers to "CDAP streams" or "DMaaP feeds". While a stream is very likely a DMaaP feed, it could be a direct stream of data being routed via HTTP too. It abstractly refers to a sequence of data leaving a publisher.
Streams have anonymous publish/subscribe semantics, which decouples the production of information from its consumption. Like the component specification, the data format specification is represented/validated against this Data Format json schema
In general, components are not aware of who they are communicating with.
Instead, components that are interested in data, subscribe to the relevant stream; components that generate data publish to the relevant stream.
There can be multiple publishers and subscribers to a stream. Streams are intended for unidirectional, streaming communication.
Streams interfaces that implement an HTTP endpoint must support POST.
Streams are split into:
Property Name | Type | Description |
---|---|---|
subscribes | JSON list | Required. List of all available stream interfaces that this component has that can be used for subscribing |
publishes | JSON list | Required. List of all stream interfaces that this component will publish onto |
Example:
"streams": { "subscribes": [{ "format": "dcae.vnf.kpi", "version": "1.0.0", "route": "/data", // for CDAP this value is not used "type": "http" }], ... }
This describes that asimov.component.kpi_anomaly
exposes an HTTP endpoint called /data
which accepts requests that have the data format of dcae.vnf.kpi
version 1.0.0
.
subscribes
Schema:
Property Name | Type | Description |
---|---|---|
format | string | Required. Data format id of the data format that is used by this interface |
version | string | Required. Data format version of the data format that is used by this interface |
route | string | Required for HTTP and data router. The HTTP route that this interface listens on |
config_key | string | Required for message_router and data router. The HTTP route that this interface listens on |
type | string | Required. Type of stream: http , message_router , data_router |
Message router subscribers are http clients rather than http services and performs a http a GET
call. Thus, message router subscribers description is structured like message router publishers and requires config_key
:
"streams": { "subscribes": [{ "format": "dcae.some-format", "version": "1.0.0", "config_key": "some_format_handle", "type": "message router" }], ... }
Data router subscribers are http or https services that handle PUT
requests from data router. Developers must provide the route
or url path/endpoint that is expected to handle data router requests. This will be used to construct the delivery url needed to register the subscriber to the provisioned feed. Developers must also provide a config_key
because there is dynamic configuration information associated with the feed that the application will need e.g. username and password. See the page on DMaaP connection objects for more details on the configuration information.
Example (not tied to the larger example):
"streams": { "subscribes": [{ "config_key": "some-sub-dr", "format": "sandbox.platform.any", "route": "/identity", "type": "data_router", "version": "0.1.0" }], ... }
Example:
"streams": { ... "publishes": [{ "format": "asimov.format.integerClassification", "version": "1.0.0", "config_key": "prediction", "type": "http" }] },
This describes that asimov.component.kpi_anomaly
publishes by making POST requests to streams that support the data format asimov.format.integerClassification
version 1.0.0
.
publishes
Schema:
Property Name | Type | Description |
---|---|---|
format | string | Required. Data format id of the data format that is used by this interface |
version | string | Required. Data format version of the data format that is used by this interface |
config_key | string | Required. The JSON key in the generated application configuration that will be used to pass the downstream component's (the subscriber's) connection information. |
type | string | Required. Type of stream: http , message router , data router |
Message router publishers are http clients of DMaap message_router. Developers must provide a config_key
because there is dynamic configuration information associated with the feed that the application needs to receive e.g. topic url, username, password. See the page on DMaaP connection objects for more details on the configuration information.
Example (not tied to the larger example):
"streams": { ... "publishes": [{ "config_key": "some-pub-mr", "format": "sandbox.platform.any", "type": "message_router", "version": "0.1.0" }] }
Data router publishers are http clients that make PUT
requests to data router. Developers must also provide a config_key
because there is dynamic configuration information associated with the feed that the application will need to receive e.g. publish url, username, password. See the page on DMaaP connection objects for more details on the configuration information.
Example (not tied to the larger example):
"streams": { ... "publishes": [{ "config_key": "some-pub-dr", "format": "sandbox.platform.any", "type": "data_router", "version": "0.1.0" }] }
Refer to this Quick Reference for a comparison of the Streams 'Publishes' and 'Subscribes' sections.
Services are split into:
Property Name | Type | Description |
---|---|---|
calls | JSON list | Required. List of all service interfaces that this component will call |
provides | JSON list | Required. List of all service interfaces that this component exposes and provides |
The JSON services/calls
is for specifying that the component relies on an HTTP(S) service---the component sends that service an HTTP request, and that service responds with an HTTP reply. An example of this is how string matching (SM) depends on the AAI Broker. SM performs a synchronous REST call to the AAI broker, providing it the VMNAME of the VNF, and the AAI Broker responds with additional details about the VNF. This dependency is expressed via services/calls
. In contrast, the output of string matching (the alerts it computes) is sent directly to policy as a fire-and-forget interface, so that is an example of a stream
.
Example:
"services": { "calls": [{ "config_key": "vnf-db", "request": { "format": "dcae.vnf.meta", "version": "1.0.0" }, "response": { "format": "dcae.vnf.kpi", "version": "1.0.0" } }], ... }
This describes that asimov.component.kpi_anomaly
will make HTTP calls to a downstream component that accepts requests of data format dcae.vnf.meta
version 1.0.0
and is expecting the response to be dcae.vnf.kpi
version 1.0.0
.
calls
Schema:
Property Name | Type | Description |
---|---|---|
request | JSON object | Required. Description of the expected request for this downstream interface |
response | JSON object | Required. Description of the expected response for this downstream interface |
config_key | string | Required. The JSON key in the generated application configuration that will be used to pass the downstream component connection information. |
The JSON object schema for both request
and response
:
Property Name | Type | Description |
---|---|---|
format | string | Required. Data format id of the data format that is used by this interface |
version | string | Required. Data format version of the data format that is used by this interface |
Example:
"services": { ... "provides": [{ "route": "/score-vnf", "request": { "format": "dcae.vnf.meta", "version": "1.0.0" }, "response": { "format": "asimov.format.integerClassification", "version": "1.0.0" } }] },
This describes that asimov.component.kpi_anomaly
provides a service interface and it is exposed on the /score-vnf
HTTP endpoint. The endpoint accepts requests that have the data format dcae.vnf.meta
version 1.0.0
and gives back a response of asimov.format.integerClassification
version 1.0.0
.
provides
Schema for a Docker component:
Property Name | Type | Description |
---|---|---|
request | JSON object | Required. Description of the expected request for this interface |
response | JSON object | Required. Description of the expected response for this interface |
route | string | Required. The HTTP route that this interface listens on |
The JSON object schema for both request
and response
:
Property Name | Type | Description |
---|---|---|
format | string | Required. Data format id of the data format that is used by this interface |
version | string | Required. Data format version of the data format that is used by this interface |
Note, for CDAP, there is a slight variation due to the way CDAP exposes services:
"provides":[ // note this is a list of JSON { "request":{ ...}, "response":{ ...}, "service_name":"name CDAP service", "service_endpoint":"greet", // E.g the URL is /services/service_name/methods/service_endpoint "verb":"GET" // GET, PUT, or POST } ]
provides
Schema for a CDAP component:
Property Name | Type | Description |
---|---|---|
request | JSON object | Required. Description of the expected request data format for this interface |
response | JSON object | Required. Description of the expected response for this interface |
service_name | string | Required. The CDAP service name (eg "Greeting") |
service_endpoint | string | Required. The CDAP service endpoint for this service_name (eg "/greet") |
verb | string | Required. 'GET', 'PUT' or 'POST' |
parameters
is where to specify the component's application configuration parameters that are not connection information.
Property Name | Type | Description |
---|---|---|
parameters | JSON array | Each entry is a parameter object |
Parameter object has the following available properties:
Property Name | Type | Description | Default |
---|---|---|---|
name | string | Required. The property name that will be used as the key in the generated config | |
value | any | Required. The default value for the given parameter | |
description | string | Required. Human-readable text describing the parameter like what its for | |
type | string | The required data type for the parameter | |
required | boolean | An optional key that declares a parameter as required (true) or not (false) | true |
constraints | array | The optional list of sequenced constraint clauses for the parameter. See below | |
entry_schema | string | The optional key that is used to declare the name of the Datatype definition for entries of set types such as the TOSCA 'list' or 'map'. Only 1 level is supported at this time | |
designer_editable | boolean | An optional key that declares a parameter to be editable by designer (true) or not (false) | |
sourced_at_deployment | boolean | An optional key that declares a parameter's value to be assigned at deployment time (true) | |
policy_editable | boolean | An optional key that declares a parameter to be editable by policy (true) or not (false) | |
policy_schema | array | The optional list of schema definitions used for policy. See below |
Example:
"parameters": [ { "name": "threshold", "value": 0.75, "description": "Probability threshold to exceed to be anomalous" } ]
Many of the parameter properties have been copied from TOSCA model property definitions and are to be used for service design composition and policy creation. See section 3.5.8 Property definition.
The property constraints
is a list of objects where each constraint object:
Property Name | Type | Description |
---|---|---|
equal | Constrains a property or parameter to a value equal to (‘=’) the value declared | |
greater_than | number | Constrains a property or parameter to a value greater than (‘>’) the value declared |
greater_or_equal | number | Constrains a property or parameter to a value greater than or equal to (‘>=’) the value declared |
less_than | number | Constrains a property or parameter to a value less than (‘<’) the value declared |
less_or_equal | number | Constrains a property or parameter to a value less than or equal to (‘<=’) the value declared |
valid_values | array | Constrains a property or parameter to a value that is in the list of declared values |
length | number | Constrains the property or parameter to a value of a given length |
min_length | number | Constrains the property or parameter to a value to a minimum length |
max_length | number | Constrains the property or parameter to a value to a maximum length |
threshold
is the configuration parameter and will get set to 0.75 when the configuration gets generated.
The property policy_schema
is a list of objects where each policy_schema object:
Property Name | Type | Description | Default |
---|---|---|---|
name | string | Required. parameter name | |
value | string | default value for the parameter | |
description | string | parameter description | |
type | enum | Required. data type of the parameter, 'string', 'number', 'boolean', 'datetime', 'list', or 'map' | |
required | boolean | is parameter required or not? | true |
constraints | array | The optional list of sequenced constraint clauses for the parameter. See above | |
entry_schema | string | The optional key that is used to declare the name of the Datatype definition for certain types. entry_schema must be defined when the type is either list or map. If the type is list and the entry type is a simple type (string, number, bookean, datetime), follow with an string to describe the entry | |
If the type is list and the entry type is a map, follow with an array to describe the keys for the entry map | |||
If the type is list and the entry type is a list, that is not currently supported | |||
If the type is map, follow with an aray to describe the keys for the map |
The Platform generates configuration for the component (based on the component spec) at deployment time. The generated application configuration will be made up of the Parameters, Streams, and Services sections, after any provisioning for streams and services. The component developer can see what this configuration will look like by reviewing the component dev command.
artifacts
contains a list of artifacts associated with this component. For Docker, this is the full path (including the registry) to the Docker image. For CDAP, this is the full path to the CDAP jar.
Property Name | Type | Description |
---|---|---|
artifacts | JSON array | Each entry is a artifact object |
artifact
Schema:
Property Name | Type | Description |
---|---|---|
uri | string | Required. Uri to the artifact, full path |
type | string | Required. docker image or jar |
Components can be added to the onboarding catalog (which first validates the component spec) by using the dcae_cli Tool. Here you can also list the components, show the contents of a component, publish the component, validate the generated configuration for the component, deploy (run) and undeploy the component. For a list of these capabilities, see Component Commands.