18
AURELIUS THINKAURELIUS.COM @mbroecheler #strangeloop 2013 Matthias Broecheler September IXX, MMXIII

Graph Computing @ Strangeloop 2013

Embed Size (px)

DESCRIPTION

Problem solving in the 21st century increasingly depends on the analysis of complex systems. Developing new drugs, understanding risk in financial networks, searching for answers in knowledge graphs, personalization and recommendation in social networks all require the analysis of systems composed of interconnected entities that exhibit complex behavior as a whole. Graph computing provides a conceptual model and practical platform for developing such analyses. This talk presents graph computing as an important component of every developer’s toolbox. We introduce the Aurelius graph cluster which is an open-source stack enabling graph computing at scale by building on distributed systems like Cassandra, HBase, and Hadoop. This stack addresses challenging problems in graph partitioning, graph query language design and graph algorithm development with solutions inspired by physics, biology and neuroscience.

Citation preview

Page 1: Graph Computing @ Strangeloop 2013

AURELIUS THINKAURELIUS.COM

@mbroecheler #strangeloop 2013

Matthias Broecheler

September IXX, MMXIII

Page 2: Graph Computing @ Strangeloop 2013

Enabling Humans

;MMI=C;NCP? G?GILS ;H> NBCHECHA

Page 3: Graph Computing @ Strangeloop 2013

What is the name of and highest degree earned by

the wife of the current president of the US?

Page 4: Graph Computing @ Strangeloop 2013

SELECT People.name, DegreeEarned.name FROM DegreeEarned INNER JOIN People ON DegreeEarned.personid=People.personid INNER JOIN Spouse ON People.personid=Spouse.wifeid INNER JOIN President ON President.personid=Spouse.husbandid INNER JOIN Country ON President.countryid=Country.countryid WHERE Country.name='USA' AND President.until=NULL ORDER BY DegreeEarned.type LIMIT 1

Page 5: Graph Computing @ Strangeloop 2013

What is the average term length of those presidents of the US that took office

in the 1980s or later?

Page 6: Graph Computing @ Strangeloop 2013

SELECT Avg(President.term_length) FROM President INNER JOIN Country ON President.countryid=Country.countryid WHERE Country.name='USA' AND President.from>1980

Page 7: Graph Computing @ Strangeloop 2013

name: George H.W. Bush

president

spouse

father

president

spouse

spouse

name: Barbara Bush

name: Laura Bush

name: George W. Bush

name: Michelle Obama

name: Barack Obama

country_name: USA

president

from: 2008 until: -

from: 2000 until: 2008

from: 1988 until: 1992

degree

degree

degree

type: school

type: college

type: graduate

name: Juris Doctor date: 1988

name: Bachelor of Arts date: 1985

name: High School Diploma date: 1981

Page 8: Graph Computing @ Strangeloop 2013

g.V('country_name', 'USA').inE('president').hasNot('until').outV .out('spouse').as('x') .out('degree')[0].as('y’).select{it.name}

president

spouse

father

president

spouse

spouse

name: Laura Bush

name: George W. Bush

name: Michelle Obama

name: Barack Obama

country_name: USA

president

from: 2008 until: -

from: 2000 until: 2008

degree

degree

degree

type: school

type: college

type: graduate

name: George H.W. Bush

name: Barbara Bush

name: Juris Doctor date: 1988

name: Bachelor of Arts date: 1985

name: High School Diploma date: 1981

from: 1988 until: 1992

Page 9: Graph Computing @ Strangeloop 2013

Enabling Humans

Complex Systems

;MMI=C;NCP? G?GILS ;H> NBCHECHA

G;>? I@ G;HS HIH C>?HNC=;F ?F?G?HNM =IHH?=N?> <S >CP?LM? CHN?L;=NCIHM

Page 10: Graph Computing @ Strangeloop 2013
Page 11: Graph Computing @ Strangeloop 2013
Page 12: Graph Computing @ Strangeloop 2013
Page 13: Graph Computing @ Strangeloop 2013

Enabling Humans to analyze

Complex Systems

;MMI=C;NCP? G?GILS ;H> NBCHECHA

G;>? I@ G;HS HIH C>?HNC=;F ?F?G?HNM =IHH?=N?> <S >CP?LM? CHN?L;=NCIHM

Page 14: Graph Computing @ Strangeloop 2013

Distributed Graph Computing at Scale TITAN.THINKAURELIUS.COM

TITAN

Page 15: Graph Computing @ Strangeloop 2013
Page 16: Graph Computing @ Strangeloop 2013

Vertex-Centric Index

NCG? CGJILN;H=? NSJ?

Page 17: Graph Computing @ Strangeloop 2013

Aurelius Graph Cluster

OLTP OLAP

Hadoop MapReduce

Analysis results back into Titan

Apache 2

g.V.label.groupCount g.v(101).out

titan.thinkaurelius.com faunus.thinkaurelius.com

Page 18: Graph Computing @ Strangeloop 2013

AURELIUS THINKAURELIUS.COM

Join Us!