Title: A Coherent and Rich PaaS with a Common Programming Model
Acronym: CoherentPaaS (Grant agreement no: 611068 – Oct 2013-Sept2016)
URL: http://coherentpaas.eu/
Abstract:
In the cloud computing landscape it has, by today, been established that there is “no one size fits all”. This has resulted in a blooming of domain specific cloud data management infrastructures specialized for different kinds of data that are able to achieve orders of magnitude better scaling and performance than traditional generic approaches. In the effort to achieve high scalability in the cloud landscape, the ACID coherence provided in traditional environments has been abandoned within and across data stores. This trend has resulted in a wide diversification of APIs and in the divergence of the programming paradigms used in different domains and for different types of data. Within a single cloud data store, such as key-value data stores or sharded SQL databases, no transactional semantics is provided at the data store level. Instead, sharding (splitting the data into many different pieces) is used to allow for linear scaling. When transactions are provided, they cannot span across shards resulting in an effective loss of transactional semantics across the full store. In addition, across data stores, there is today no notion of coherence for data. Consequently, accessing and managing large amounts of data is becoming a major obstacle to developing new cloud applications and services with correct semantics, requiring tremendous programming effort and expertise.
CoherentPaaS addresses this issue in the cloud PaaS landscape by developing a PaaS that: i) Incorporates a rich and diverse set of cloud data management technologies, including no SQL data stores, such as key-value data stores and graph databases, SQL data stores, such as in-memory and column-oriented databases, hybrid systems, such as SQL engines on top on key-value data stores, and complex event processing data management systems, ii) uses a common query language to unify the programming models of all systems under a single paradigm, iii) Provide holistic coherence across data stores using a scalable, transactional management system.
The Role of DKMS lab: participant
DKMS is involved mainly involved in the incorporation of the document cloud data stores in the CoherentPaaS and also in the scalable transactional processing. In this scope, a well-known document data store, MongoDB, will be extended and enhanced to support the transactional model of the holistic transactions, providing snapshot isolation via the multi-version concurrency control in order to expose versioned objectContact Points:s and to export an appropriate interface and correct semantics so as to allow higher-layers to properly access internal key-value pairs in the context of concurrent transactions, maintaining proper recoverability at the same time. DKMS will be also involved in the definition of the common query language that will support inter-exchangeability between different cloud data stores and, in this scope, a local query engine will be implemented to undertake the task of integrating MongoDB with CoherentPaaS. A local transactional manager will be also developed along with an appropriate recovery protocol and versioned garbage collector. DKMS will assist in the design of the X-Ray monitoring that will provide fine-grain analysis of cloud applications deployed on CoherentPaaS, exploiting relevant statistics of their operation through probes that will be exposed via standardized interfaces in order to achieve profiling instrumentation. Finally, DKMS will lead all the dissemination activities of the project and will organize two open workshops for the key players in the cloud landscape.
Contact Points: Marta Patiño mpatino@fi.upm.es