Within the first open source framework for quality-aware DevOps for Big Data applications – DICE, TOSCA provides the perfect foundation for bridging the world of off-line application modelling and the running instances of this application. Its extensibility enabled automating deployments of complex data-intensive applications.
The DICE project is helping SMEs achieve Big DataOps in no time – more specifically, the project’s key aim was to provide a DevOps methodology and toolkit to help architects and developers of Big Data applications to quickly and easily model, evaluate, deploy and improve their data intensive applications.
In the arsenal of the DICE toolset practitioners find:
- a UML profile and metamodel that enriches the models with expressiveness related to describing non-functional requirements related to data streaming, batch processing and storage,
- the tools for transforming models into a formal language ready for optimisation, as well as subsequent cloud application orchestration, monitoring, anomaly detection and other runtime feats.
February 2018 was the month when the H2020 project DICE officially ended. In March, the project received high praises and favourable remarks from the European Commission reviewers responsible for evaluating the project.
Within the whole DICE stack, TOSCA plays a pivotal role as the language for cloud blueprints. On the one hand, it lets the users quickly build blueprints of minimal working platforms, such as a simple HDFS cluster. The same language is comfortable for representing a complex stack of an application that ingest streams of data using Kafka, performs stream analytics in Storm, stores the results in Cassandra, and batch-process it using Spark. DICE provides a TOSCA profile for simple blueprinting of applications that mix popular open source Big Data components with Docker containers or OSv unikernels.
The toolset is powered by the Cloudify orchestrator, thus the blueprints are in Cloudify DSL that is inspired by TOSCA. Nevertheless, this demonstrates the TOSCA’s strength of solving general use cases of multi-cloud application deployments, while also letting us build a powerful profile that is focused on the Big Data use case.
Matej Artač, XLAB
Damian A. Tamburri, PMI