| Table of Contents |
|---|
DCAE platform being microservice centric, the services themselves can be build independently and integrated with DCAE with minimal work. However as ONAP itself requiring carrier grade level, all components has strict S3P goals and LF process to comply under DCAE project.
...
Once the main functionality is build and components is executable as docker container/jar - it should be integrated with DCAE platform to enable dynamic instantiation and configuration management by DCAE controller.
Build component spec and validate using dcae_cli (Design support)
Every component to be onboarded into DCAE, should prepare a component spec (a.k.a spec) - which is meta data represented in json describing the component configuration model. Details on spec creation can be found here - https://onap.readthedocs.io/en/beijing/submodules/dcaegen2.git/docs/sections/components/component-specification/component-specification.html?highlight=component%20spec
The dcae_cli tools enables MS owner to validate the component spec and data_Formats created and also "test" the deployment of MS itself. This allows components to mimic the configuration returned from ConfigBinding Service as expected on typical cloudify based deployment by DCAE platform (Documentation on spec format and tool usage can be found under https://onap.readthedocs.io/en/beijing/submodules/dcaegen2.git/docs/sections/components/dcae-cli/commands.html?highlight=dcae_cli )
SDC Integration (Design support)
This has to be coordinated with SDC team esp if components has dynamic policy and required CLAMP/Policy integration. This step allows Tosca models to be generated via DCAE-DS/Tosca_Lab tool that will be used in service composition and distribution to Policy/CLAMP and DCAE.
Blueprint creation/test (Test)
If the component are onboarded via SDC flow, the blueprints will be generated by TOSCAlab tools contained with DCAE-DS in SDC. Alternatively the components developer can hand-craft the blueprints and test. The blueprint created manually if required to included part of DCAE static deployment, then blueprint itself should be templatized and added into this repo - https://gerrit.onap.org/r/#/admin/projects/dcaegen2/platform/blueprints. The blueprint can tested using "cfy" commands (Ms Deployment through Cloudify CLI) pointed to correct DCAE cloudify server.
ConfigBindingService Integration (Code enhancement required)
All configuration (identified in component spec - parameters) itself are stored under CONSUL when blueprint gets executed part of component deployment. Some configuration such as topics are dynamic and may not be known for component owner to include in the docker image. Hence an important requirement for all onboarding services is to invoke the configbindingservice api during docker init script process to fetch the required application config.
...
For python - there are libraries available which can be used to invoke this api. For java - sample code and configuration returned can be found here MicroServices Onboarding
Notification mechanism (Code enhancement required)
Support for notification mechanism integration is required within MS container/package. The component spec owner will declare in the spec the script that platform should invoke when a configuration change is received (eg. from Policy) This script should be contained within the MS container/package and have necessary logic to invoke CBS API and cascade the information for application processing. For components not expecting dynamic changes, this step is optional.
S3P Support (Code enhancement required)
If the component is stateful, it should persist its state on external store (eg. pg, redis) to allow support for scaling and resiliency. This should be important design criteria for the component. If the component fails to meet S3P goals, it will not be release candidate.
...