Lionel Brunie

Similarity-based operators and query optimization for multimedia database systems

The many successful research results in the domain of computer vision have made similarity based data retrieval techniques a promising approach. As a result, the integration of similarity based retrieval techniques of multimedia data into DBMSs is currently an active research issue. We first illustrate the importance of similarity based operations. Then, we present our image data repository model that supports similarity based operations conveniently under an object-relational database paradigm. Furthermore, we present novel similarity based operators on image tables and study their properties. Finally, based on the properties of the operators identified, we derive algebraic rules that are useful for similarity based query optimization and we introduce a cost model for an implementation of one of the major similarity based operators.

Control strategies for complex relational query processing in shared nothing systems

In this paper, we present an original and complete methodology for supervising relational query processing in shared nothing systems. A new control mechanism is introduced which allows the detection and the correction of optimizer estimation errors and load imbalance. We especially focus on the management of intraprocessor communication and on the overlapping of communication and computation. Performance evaluations on an hypercube and a grid interconnection machine show the efficiency and the robustness of the proposed methods.

Semantic Based Prefetching in News-on-Demand Video Servers

In video databases, a video document has two abstractions. The high level abstraction corresponds to the view in which the contents of that video document are seen by end users, and the low level abstraction corresponds to the physical organization of that video document. Due to the huge size of continuous data, reducing I/O has become a key issue. The latter has been mostly addressed by developing appropriate buffering techniques. In addition, prefetching techniques play a major role to meet the video data requirements. In this paper, we propose a novel prefetching strategy based not only on run-time information (objects access frequencies for example) but also on knowledge about clips structures. The proposed technique merges the two views of a video document to trigger prefetching at the video server level. Simulation experiments for a News-on-Demand application performed on different request scenarios show an improvement of about 18% in the buffer hit-rate with respect, first to the available buffer size and second to the request arrival rate.

A PC-NOW Based Parallel Extension for a Sequential DBMS

In this paper we study the use of networks of PCs to handle the parallel execution of relational database queries. This approach is based on a parallel extension, called parallel relational query evaluator, working in a coupled mode with a sequential DBMS. We present a detailed arc hitecture of the parallel query evaluator and introduce Enkidu, the effiient Java-based prototype that has been build according to our concepts. We expose a set of measurements, conducted over Enkidu, and highlighting its performances.We finally discuss the interest and viability of the concept of parallel extension in the context of relational databases and in the wider context of high performance computing.

Execution analysis of DSM applications: a distributed and scalable approach

In the last five years, Distributed Shared Memory (DSM) systems have received increasing attention. Indeed, by releasing the programmer from the management of inter-process communications, they offer a very intuitive and easyto-use programming paradigm. In compensation, such systems often appear, from the programmer point of view, as a “black box” since no information about the actual communications is available. Consequently, in the absence of visualization and monitoring tools, optimizing, debugging or evaluating the performance of DSM applications is very difficult. In that framework, this paper proposes an original monitoring model based on two new concepts: rneta-objects and event manager processes. This model constitutes the basis of an actual monitoring system, called DOSMOS-Trace, that has been designed and implemented to monitor applications developed on top of the DOSMOS DSM systeml. This monitoring environment is analysed in terms of functionalities, protocols and user interface. Experiments show the efficiency and the robustness of the underlying model as well as the pertinence, for the programmer, of such a monitoring tool.

Software architectures for collaborative proxies in wide area information systems

With the fantastic development of network technologies (LAN, Intranet, Internet), information systems are every day more interconnected. However information system interconnection poses several issues: an evident issue is the necessity to use or develop a common representation of the target application: in this framework lie the works concerning ontologies, wrappers, brokers, collaborating query agents, etc. Another important issue is the performance of the global interconnected system: indeed interconnecting information systems increases the data servers load, the network load and may also cause query answer delays. Finally, and this may be the main key issue, too much information is no information: when the quantity of data becomes too important, users need help and tools to find their way in the information space. In this framework, we propose research directions based on collaborative proxy architectures to enhance the global efficiency of interconnected information systems (e.g., the WWW) distributed over a WAN, to provide users with adaptive search indexes and to build virtual communities.

Encapsulation of Intra-operation Parallelism in a Parallel Match Operator

This paper presents a new concept for relational parallel query optimization. We propose to encapsulate the intra-operation parallelism in a parallel match operator realizing most binary relational operations. Our approach integrates the optimization of this operator: a comparative study of the three main match algorithms — Nested Loops, Sort-Merge and Hash — is presented for shared-nothing architecture. It shows that database systems could profit by the implementation of all these algorithms. Futhermore, our match optimizer can be used by the query parallelizer in order to determine an optimal parallel execution strategy for each binary relational operation.

High performance distributed objects for cluster computing

With the development of cluster computing, distributed shared memory (DSM) systems have a new role to play by providing ease of programming a shared space combined with the performance of a high speed network. In this paper, we show that, by combining a high performance network with low latency communication protocols and multithreading, DSM systems on top of clusters can be an efficient alternative programming model to message passing. We present the DOSMOS system and experiments made on a Myrinet network based cluster. The raw performance of DOSMOS and results obtained with several kinds of applications are also described.

Technology transfer within the ProHPC TTN at ENS Lyon

This article is devoted to the description of our activities related to transferring the HPCN technology to SMEs. This work is performed in the framework of the French TTN ProHPC which we briefly describe in the beginning of the paper. Then we move to a more technical description of our activities, which include seven projects within the awareness campaign PHPC-ACRA and the participation to a Best Practice prototype.

Technology Transfer within the ProHPC TTN at ENS Lyon

This article is devoted to the description of our activities related to transferring the HPCN technology to SMEs. This work is performed in the framework of the French TTN ProHPC which we briefly describe in the beginning of the paper. Then we move to a more technical description of our activities, which include four projects within the awareness campaign PHPC-ACRA and the participation to a Best Practice prototype.