- Cloud for Data-Driven Policy Management
- Contact Us
- Join our community
- Login
Reusable Models & Analytical Tools: Design and Open Specification
Policy Cloud has published Reusable Models & Analytical Tools: Design and Open Specification 1. Authored by experts from Policy Cloud partner organisations ATOS, IBM, LeanXcale, the University of Piraeus Research Centre (UPRC), and OKYS, the report describes the work completed by the end of month 8.
As the first deliverable of WP4, Reusable Models & Analytical tools, the report specifies the architecture of the data acquisition and analytics layer as well as the cloud platform and software tools currently planned to be used for the implementation of the Policy Cloud demonstration platform.
Data Acquisition and Analytics Layer
The data acquisition and analytics layer is responsible for:
- Registration of data sources of various kinds and properties
- Ingest processing of the data sources
- Registration of data analytic and transformation tools
- The built-in tools for data cleaning, interoperability, and analysis.
- Big data storage
In essence, the Data Acquisition and Analytics Layer provides an extensible framework for data source and analytic tools, controlling the full data pass from the data sources through filtering, transformation and initial analytic, to hot storage and then to cold storage while enabling deeper analytics by the registered analytic tools on the ingested data at any time.
Cloud Platform and Software Tools
All software components of Policy CLoud will be deployed on a Kubernetes cluster which will be the underlying application management platform. Kubernetes is the leading open source container management platform.
All analytic functions in Policy Cloud will be deployed as OpenWhisk functions, with appropriate triggers and activation rules, addressing the extendibility and reusability requirements.
The Spark cluster will be used in Policy Cloud in conjunction with the LeanXcale database to provide seamless analytic on hot and cold data at rest, where the Spark SQL with optimization exploited from the BigDataStack EU project will be used for queries on the colder data in the object storage.