CoherentPaaS: A Coherent and Rich PaaS with a Common Programming Model

Name: 
Ricardo Jimenez-Peris
Organisation: 
Universidad Politecnica Madrid
Topics recommended for the 2016-2017 Work Programme: 

Data center disaster tolerance: Providing cloud infrastructure that is able to survive disasters destroying or rendering unavailable whole data centers in order to provide highly available cloud infrastructure so businesses can survive despite this kind of disasters (as it happened recently with the hurricanes in US and the floods in NY). Scalable cloud platforms for machine-to-machine and IoT applications: M2M and IoT applications are ideal to run in the cloud. However, they are pushing the needs for scalability by 1-2 orders of magnitude compared to current cloud applications what will require new cloud platforms able to address these scalability challenges.

Projects major results: 

Currently, companies are using a combination of cloud SQL and NoSQL technologies to deal with their data management needs. This is what is called polyglot persistence. However, polyglot persistence is bringing new problems. First, there is no consistency across different data stores. Second, queries involving multiple data stores imply to program the query manually and optimize it manually as well. CoherentPaaS is solving the issues of polyglot persistence by first integrating an ultra-scalable transactional processing that provides holistic transactions across all data stores NoSQL (key-value data store, graph DB, document-oriented data store) and SQL. This provides full ACID transactions for updates across data stores and therefore, guaranteeing full data consistency in the advent of failures and concurrent accesses. Secondly, CoherentPaaS is creating a query engine enabling to perform SQL queries to enable to correlate and aggregate data across data stores. This will enable for instance to perform joins between different data stores. In order to keep the power of the APIs/query languages of each data store, the SQL enables to embed subqueries written in the APIs/query language of the underlying data stores. In this way, CoherentPaaS gets the best of the two worlds, the power of local query languages/APIs and the ease of use of SQL.

Potential exploitation strategy: 

A startup, LeanXcale, has been created by two partners of the project for commercialising some of the core results of the project. There is a large fraction of the consortium partners that are data store providers that will integrate their data stores with the ultra-scalable transactional processing. The exploitation plan lies in signing bilateral commercial agreements between LeanXcale and each of these partners to commercialise the integrated systems. The consortium also brings end user partners that are potential clients of the resulting platform. The startup is being constitute legally and will be created by end of September 2014.

An update since the last Concertation meeting (March 2014): 

CoherentPaaS has partnered with the LeanBigData project also led by UPM on an ultra-scalable big data platform. These two projects will integrate their results what will result in a cloud data management platform providing not only an integrated data management platform for OLTP with SQL and NoSQL data stores, but also with OLAP capabilities to perform analytics on top of it.