Workshop : AGERE 2012

Applications running over decentralized systems, distribute their computation on nodes/agents,  which exchange data and services through messages. In many cases, the provenance of the data or service is not relevant, and applications can be optimized by choosing the most efficient solution  to obtain them. We introduce a framework which allows messages with intensional destination, which can be seen as restricted mobile agents, specifying the desired service but not the exact node that carries it, leaving to the system the task of evaluating the extensional destination, that is an explicit address for that service. The intensional destinations are defined using queries that are evaluated by other agents while routing. We introduce the Questlog language, which allows to reformulate queries, and express complex strategies to pull distributed data. In addition, intensional addresses offer persistency to dynamic systems with nodes/agents leaving the system. We demonstrate the approach with examples taken from sensor networks, and show some experimental results on the QuestMonitor platform.

Conference : BDA 2011

Abstract :

We present an approach based on peer-to-peer overlays which allows to distribute seamlessly data centric applications defined by queries over a centralized database. We consider applications in which the users have access to views, which contain horizontal fragments of the data of interest to them. The peer-to-peer overlays are defined by simple declarative programs in the Netlog language. The communication relies on implicit addresses, which can be evaluated on the fly, and ensure persistence of data in messages. We demonstrate the technique on a multiplayer online game, written in SQL, with players who connect to a mobile ad hoc network through their portable devices. The overlay is defined by a combination of an ad hoc routing protocol, DSDV, together with a DHT. The application runs on the QuestMonitor platform, which allows to monitor the communication between peers, the evolution of the local data stores, as well as the execution of the declarative code.

Conferences : BDA 2012, IDEAS 2012

Abstract :
This paper describes the QOL approach to optimize distributed queries by learning. It is well-adapted to social systems (e.g. games, social networks, sharing), where data are pushed or pulled with incomplete knowledge in a dynamic environment. The contribution of this work is twofold. It first concerns the integration of the Case Based Reasoning (CBR) paradigm in query processing, providing a way to optimize queries when there is no prior knowledge on queried data sources and certainly no related metadata such as data statistics. Our approach optimizes queries using cases generated from the evaluation of similar past queries. A query case comprises: (i) the query, (ii) the query plan and (iii) the measures (computational resources consumed) of the query plan. The second aspect of the work concerns the way the CBR process interacts with the query plan generation process. This process uses classical heuristics and makes decisions randomly (e.g. when there is no statistics for join ordering and selection of algorithms, routing protocols); It also (re)uses cases (existing query plans) for similar queries parts, improving the query optimization and evaluation efficiency.

Conferences : BDA 2012, IDEAS 2012

Abstract :
The UBIQUEST project proposes a high level programming abstraction for rapid prototyping of heterogeneous and distributed applications in a dynamic environment. Such an environment is perceived as a distributed database and the applications interact through declarative queries including declarative networking programs (e.g. routing) and/or specific data-oriented distributed algorithms (e.g. distributed join). Rule programs are specified using Datalog-based languages, while data oriented manipulations are specified using an SQL-Like language. Case-Based Reasoning is used for optimization of distributed queries when as there is no prior knowledge on data (sources) in networking applications, and certainly no related metadata such as data statistics. An UBIQUEST system is therefore well adapted to social systems (e.g. games, social networks, sharing), where data are pushed or pulled without knowledge or with incomplete knowledge.

Conference : DATA 2012

Abstract :
The paper introduces our vision for rapid prototyping of heterogeneous and distributed applications. It abstracts a network as a large distributed database providing a unified view of "objects" handled in networks and applications. The applications interact through declarative queries including  declarative networking programs (e.g. routing) and/or specific data-oriented distributed algorithms (e.g. distributed join). Case-Based Reasoning is used for optimization of distributed queries by learning when there is no prior knowledge on queried data sources and no related metadata such as data statistics.