In previous blog posts, we have mentioned that RADON aims to provide a unified DevOps experience allowing to employ serverless Function as a Service (FaaS) technology in modern software system engineering. In order to achieve this, RADON introduces a novel modeling approach and a set of standardized, reusable model components for developing and orchestrating the deployment and execution of microservices, FaaS-based serverless applications, and data processing pipelines.
In the Public Deliverable ‘4.3 RADON Models I’, our team presented an initial version of RADON modeling approach consisting of abstract and deployable entity modeling layers that comprise a set of abstract and technology-specific modeling constructs addressing the RADON modeling challenges.
Michael Wurster and Vladimir Yusupov from the University of Stuttgart, Damian Tamburri from Tilburg University and Lulai Zhu from the Imperial College London, outline the modeling specifics with respect to the challenges in the context of RADON together with the detailed description of models in the D4.3 companion document. Their findings served as a basis for the final RADON Models deliverable.
In addition, RADON models utilize and extend the TOSCA language for emerging compute contexts. With the formula “emerging compute contexts”, RADON intends to target:
- serverless and FaaS-enabled orchestration
- microservice orchestration
- data pipeline orchestration.
As the authors point out, ‘TOSCA is usable and adaptable for the aforementioned emerging compute contexts’. Through this work, the RADON team has set the following objectives:
- Define best practices for the usage of TOSCA in such contexts
- Publish a first working profile providing reusable TOSCA types for such contexts
- Provide a series of abstractions that may consolidate or compound the usage of TOSCA
Last but not least, the main achievements described in this deliverable include:
- Introduction of two modeling layers that can be used to model on abstract level as well as on concrete level to produce deployable blueprints
- Presentation of RADON’s type hierarchy and the intended use
- Detailed discussion of different modeling styles in the FaaS context by employing different function triggering semantics using the notion of “invocable” and “standalone” functions
- Presentation of an initial approach to model application components of different granularity, e.g., modeling of FaaS-based microservices with traditional components
- Introduction of a first step towards the modeling and orchestration of data pipelines using TOSCA to filter, transform, or analyze streams of data
- Implementation of RADON’s Template Library in the form of a publicly maintained GitHub repository
- Definition of initial reusable types as defined in detail in the companion document to cover the reference application level technologies agreed upon the consortium
All models analysed in the document are publicly-available in the so-called “RADON Particles” repository. In addition, a respective companion document covers the detailed presentation of these “RADON Particles” in the form of TOSCA type specifications.
You can read the Public Deliverable ‘4.3 RADON Models I’ here.