10
Triple Storage

Triple Storage. Copyright 2006 by CEBT Triple(RDF) Storages A triple store is designed to store and retrieve identities that are constructed from

Embed Size (px)

Citation preview

Page 1: Triple Storage. Copyright  2006 by CEBT Triple(RDF) Storages  A triple store is designed to store and retrieve identities that are constructed from

Triple Storage

Page 2: Triple Storage. Copyright  2006 by CEBT Triple(RDF) Storages  A triple store is designed to store and retrieve identities that are constructed from

Copyright 2006 by CEBTCenter for E-Business Technology

Triple(RDF) Storages

A triple store is designed to store and retrieve identities that are constructed from triplex collections of strings

These triplex collections represent a Subject-Predicate-Object relationship

Each RDF statement is stored as a single row in a three column table.

IDS Lab. Seminar - 2

Page 3: Triple Storage. Copyright  2006 by CEBT Triple(RDF) Storages  A triple store is designed to store and retrieve identities that are constructed from

Copyright 2006 by CEBTCenter for E-Business Technology

Features

Jena Support for multiple reasoning engines

– OWL-DL, OWL-Lite, RDF Scheme

Jena’s fundamental class for users is the Model, an API for dealing with a set of RDF triples.

A Model can be created from the file system, existing RDBMS such as MySQL or PostgreSQL

Sesame Supports the SeRQL query language

Supports file system, In-memory and existing table-based RDBMS such as MySQL and PostgreSQL

Fair performance with systems of up to 70M triples.

IDS Lab. Seminar - 3

Page 4: Triple Storage. Copyright  2006 by CEBT Triple(RDF) Storages  A triple store is designed to store and retrieve identities that are constructed from

Copyright 2006 by CEBTCenter for E-Business Technology

Feature

Mulgara Designed for extremely large graphs

Good performance with stores of 160M triples

Uses a proprietary storage (not existing RDBMS)

3Store 3store is a core C library that uses MySQL, available under the GPL license

Successful applications handling 100M triples.

Provides C language library

Uses MySql as Storage

IDS Lab. Seminar - 4

Page 5: Triple Storage. Copyright  2006 by CEBT Triple(RDF) Storages  A triple store is designed to store and retrieve identities that are constructed from

Copyright 2006 by CEBTCenter for E-Business Technology

Features

RDF Gateway Commercial Engine (free for evaluation)

Complete Web server that manages a built-in RDF Store

Proprietary RDFQL language (SPARQL is also supported)

Access control using NT user, Groups

Transaction Supported

AllegroGraph Fair performance with systems of up to billions of triples

Single threaded server based RDF store.

Multi volumn Support

IDS Lab. Seminar - 5

Page 6: Triple Storage. Copyright  2006 by CEBT Triple(RDF) Storages  A triple store is designed to store and retrieve identities that are constructed from

Copyright 2006 by CEBTCenter for E-Business Technology

Comparison of Storages

Engine Features

IDS Lab. Seminar - 6

Engines Multi-VolumnManagement Tools

View Support ShadowingTransaction ACID

Jena w/ MySQL n/a YES n/a YES n/a

Jena w/ Post-Gres

n/a YES n/a YES YES

Jena w/ InnoDB YES YES n/a YES YES

AllegroGraph n/a Few n/a n/a YES

Sesame n/a Some n/a n/a YES

Mulgara n/a No n/a n/a YES

RDF Gateway n/a Some Yes n/a YES

Page 7: Triple Storage. Copyright  2006 by CEBT Triple(RDF) Storages  A triple store is designed to store and retrieve identities that are constructed from

Copyright 2006 by CEBTCenter for E-Business Technology

Comparison of Storages

Server, API features

IDS Lab. Seminar - 7

System Creation QueryUpdate Query

ReificationCommunity Support

Reasoning

Jena YES Sparql No Yes Yes OWL-DL

Sesame YES SerQL No No Yes RDFS

Mulgara YES Itql YES No Small OWL-Lite

3Store No Sparql No No Small ?

AllegroGraph YES Sparql No Yes Small OWL

RDF Gate-way

YES Proprietary no No Small RDFSRDFS

Page 8: Triple Storage. Copyright  2006 by CEBT Triple(RDF) Storages  A triple store is designed to store and retrieve identities that are constructed from

Copyright 2006 by CEBTCenter for E-Business Technology

Comparison of Storages

Performance

IDS Lab. Seminar - 8

System Scalability Loading RDF

Jena 1.7 billion 1900 ms

Sesame 0.07 billion 1887 ms

Mulgara 0.5 billion /

3Store 0.1 billion 1908 ms

AllegroGraph 1.0 billion /

RDF Gateway 0.26 billion /

Page 9: Triple Storage. Copyright  2006 by CEBT Triple(RDF) Storages  A triple store is designed to store and retrieve identities that are constructed from

Copyright 2006 by CEBTCenter for E-Business Technology

Conclusion

Jena has reasonable scalability and performance, easy to use, and used in many projects

MySQL 3 performs the most quickly in general as a Jena Store

IDS Lab. Seminar - 9

Page 10: Triple Storage. Copyright  2006 by CEBT Triple(RDF) Storages  A triple store is designed to store and retrieve identities that are constructed from

Copyright 2006 by CEBTCenter for E-Business Technology

References

Features

RDF Storage and Retrieval Systems, Handbook of Ontologies, Springer, 2008.

Features of an Enterprise-ready Triple Store, IBM, 2006

Provenance Store Evaluation, Pacific Northwest National Lab, 2008

Scalability, Benchmark

http://esw.w3.org/topic/LargeTripleStores

http://esw.w3.org/topic/RdfStoreBenchmarking

Scalability Report on Triple Store Applications, 2004

An Evaluation of Triple-Store, ESG-CET, 2008

Sciencific Workflow Provenance Metadata Management Using an RDBMS-based RDF Store, 2007

Triple Store Schemes

http://www.w3.org/2001/sw/Europe/reports/scalable_rdbms_mapping_report/#sec-impl

IDS Lab. Seminar - 10