Minh-Duc Pham

The LDBC Social Network Benchmark: Interactive Workload

The Linked Data Benchmark Council (LDBC) is now two years underway and has gathered strong industrial participation for its mission to establish benchmarks, and benchmarking practices for evaluating graph data management systems. The LDBC introduced a new choke-point driven methodology for developing benchmark workloads, which combines user input with input from expert systems architects, which we outline. This paper describes the LDBC Social Network Benchmark (SNB), and presents database benchmarking innovation in terms of graph query functionality tested, correlated graph generation techniques, as well as a scalable benchmark driver on a workload with complex graph dependencies. SNB has three query workloads under development: Interactive, Business Intelligence, and Graph Algorithms. We describe the SNB Interactive Workload in detail and illustrate the workload with some early results, as well as the goals for the two other workloads.

Linked stream data processing engines: facts and figures

Linked Stream Data, i.e., the RDF data model extended for representing stream data generated from sensors social network applications, is gaining popularity. This has motivated considerable work on developing corresponding data models associated with processing engines. However, current implemented engines have not been thoroughly evaluated to assess their capabilities. For reasonable systematic evaluations, in this work we propose a novel, customizable evaluation framework and a corresponding methodology for realistic data generation, system testing, and result analysis. Based on this evaluation environment, extensive experiments have been conducted in order to compare the state-of-the-art LSD engines wrt. qualitative and quantitative properties, taking into account the underlying principles of stream processing. Consequently, we provide a detailed analysis of the experimental outcomes that reveal useful findings for improving current and future engines.

S3G2: A Scalable Structure-Correlated Social Graph Generator

Benchmarking graph-oriented database workloads and graph-oriented database systems is increasingly becoming relevant in analytical Big Data tasks, such as social network analysis. In graph data, structure is not mainly found inside the nodes, but especially in the way nodes happen to be connected, i.e. structural correlations. Because such structural correlations determine join fan-outs experienced by graph analysis algorithms and graph query executors, they are an essential, yet typically neglected, ingredient of synthetic graph generators. To address this, we present S3G2: a Scalable Structure-correlated Social Graph Generator. This graph generator creates a synthetic social graph, containing non-uniform value distributions and structural correlations, which is intended as test data for scalable graph analysis algorithms and graph database systems. We generalize the problem by decomposing correlated graph generation in multiple passes that each focus on one so-called correlation dimension; each of which can be mapped to a MapReduce task. We show that S3G2 can generate social graphs that (i) share well-known graph connectivity characteristics typically found in real social graphs (ii) contain certain plausible structural correlations that influence the performance of graph analysis algorithms and queries, and (iii) can be quickly generated at huge sizes on common cluster hardware.

Self-organizing structured RDF in MonetDB

The semantic web uses RDF as its data model, providing ultimate flexibility for users to represent and evolve data without need of a schema. Yet, this flexibility poses challenges in implementing efficient RDF stores, leading from plans with very many self-joins to a triple table, difficulties to optimize these, and a lack of data locality since without a notion of multi-attribute data structure, clustered indexing opportunities are lost. Apart from performance issues, users of huge RDF graphs often have problems formulating queries as they lack any system-supported notion of the structure in the data. In this research, we exploit the observation that real RDF data, while not as regularly structured as relational data, still has the great majority of triples conforming to regular patterns. We conjecture that a system that would recognize this structure automatically would both allow RDF stores to become more efficient and also easier to use. Concretely, we propose to derive self-organizing RDF that stores data in PSO format in such a way that the regular parts of the data physically correspond to relational columnar storage; and propose RDFscan/RDFjoin algorithms that compute star-patterns over these without wasting effort in self-joins. These regular parts, i.e. tables, are identified on ingestion by a schema discovery algorithm - as such users will gain an SQL view of the regular part of the RDF data. This research aims to produce a state-of-the-art SPARQL frontend for MonetDB as a by-product, and we already present some preliminary results on this platform.

Exploiting Emergent Schemas to Make RDF Systems More Efficient

We build on our earlier finding that more than 95 % of the triples in actual RDF triple graphs have a remarkably tabular structure, whose schema does not necessarily follow from explicit metadata such as ontologies, but for which an RDF store can automatically derive by looking at the data using so-called “emergent schema” detection techniques. In this paper we investigate how computers and in particular RDF stores can take advantage from this emergent schema to more compactly store RDF data and more efficiently optimize and execute SPARQL queries. To this end, we contribute techniques for efficient emergent schema aware RDF storage and new query operator algorithms for emergent schema aware scans and joins. In all, these techniques allow RDF schema processors fully catch up with relational database techniques in terms of rich physical database design options and efficiency, without requiring a rigid upfront schema structure definition.

Advances in Large-Scale RDF Data Management

One of the prime goals of the LOD2 project is improving the performance and scalability of RDF storage solutions so that the increasing amount of Linked Open Data (LOD) can be efficiently managed. Virtuoso has been chosen as the basic RDF store for the LOD2 project, and during the project it has been significantly improved by incorporating advanced relational database techniques from MonetDB and Vectorwise, turning it into a compressed column store with vectored execution. This has reduced the performance gap (“RDF tax”) between Virtuoso’s SQL and SPARQL query performance in a way that still respects the “schema-last” nature of RDF. However, by lacking schema information, RDF database systems such as Virtuoso still cannot use advanced relational storage optimizations such as table partitioning or clustered indexes and have to execute SPARQL queries with many self-joins to a triple table, which leads to more join effort than needed in SQL systems. In this chapter, we first discuss the new column store techniques applied to Virtuoso, the enhancements in its cluster parallel version, and show its performance using the popular BSBM benchmark at the unsurpassed scale of 150 billion triples. We finally describe ongoing work in deriving an “emergent” relational schema from RDF data, which can help to close the performance gap between relational-based and RDF-based storage solutions.