Graphs are well-suited for many use cases to express and process complex relationships among entities in enterprise and social contexts. Fueled by the growing interest in graphs, there are various graph databases and processing systems that dot the graph landscape. JanusGraph is a community-driven project that continues the legacy of Titan, a pioneer of open source graph databases. JanusGraph is a scalable graph database optimized for large scale transactional and analytical graph processing. In the session, we will introduce JanusGraph, which features full integration with the Apache TinkerPop graph stack. We will discuss JanusGraph's optimized storage model that relies on HBase for fast graph transversal and processing. by Jason Plurad and Jing Chen He of IBM

1. Jing Chen He • jinghe@us.ibm.com • Apache HBase PMC • JanusGraph TSC
Jason Plurad • pluradj@us.ibm.com • Apache TinkerPop PMC • JanusGraph TSC
HBaseCon West 2017 • June 12, 2017
Community-Driven Graphs with
JanusGraph

2. Agenda
Property Graphs
Graph Community
Introduction to JanusGraph
JanusGraph with HBase
2 #HBaseCon

3. Graph
 Born for relationship!
 Intuitive modeling
 Expressive querying
 Native analysis
3 #HBaseCon
https://tinkerpop.apache.org/docs/3.2.4/reference/#intro

4. Graph Data Use Cases
 Social network analysis
 Configuration management database
 Master data management
 Recommendation engines
 Knowledge graphs
 Internet of things
 Cybersecurity attack analysis
4 #HBaseCon

5. Apache TinkerPop
 Open source, vendor-agnostic,
graph computing framework
 Gremlin graph traversal language
5
Apache TinkerPop™
Maintainer Apache
Software
Foundation
License Apache
Latest Release 3.2.4
February 2017
https://tinkerpop.apache.org
#HBaseCon

6. Gremlin Graph Traversal Language
6 #HBaseCon
https://tinkerpop.apache.org/gremlin.html

7. TinkerPop Stack
7 #HBaseCon
https://tinkerpop.apache.org/docs/3.2.4/reference/#_graph_system_integration

8. Graph Landscape
8 #HBaseCon
https://tinkerpop.apache.org/gremlin.html#oltp-and-olap-traversals

9.  Scalable graph database distributed on
multi-machine clusters with pluggable storage
and indexing
 Fully-compliant with Apache TinkerPop graph
computing framework
 Vendor-neutral, open community with
open governance
– Founding members: Expero, Google, GRAKN.AI,
Hortonworks, IBM
– Latest members: Amazon, Netflix, Orchestral
Developments, Uber
9
JanusGraph™
Maintainer Linux
Foundation
License Apache
Latest
Release
0.1.0
April 2017
https://janusgraph.org
#HBaseCon

10. 10 #HBaseCon
Architecture
Google Cloud Bigtable
http://docs.janusgraph.org/latest/arch-overview.html

11. 11 #HBaseCon
Storage Model
http://docs.janusgraph.org/latest/data-model.html#_janusgraph_data_layout

12. 12 #HBaseCon
Storage Model
http://docs.janusgraph.org/latest/data-model.html#_individual_edge_layout

13. 13 #HBaseCon
with HBase
 HBase – Perfect Storage Backend for JanusGraph
Big enough for your biggest graph!
The storage model
Read and write speed
Scalability and partitioning
Strong consistency
Tight integration with Hadoop Ecosystem
Great open community!
http://docs.janusgraph.org/latest/hbase.html

14. 14 #HBaseCon
with HBase
 HBase – Perfect Storage Backend for JanusGraph
Simple configuration!
 conf/janusgraph-hbase-solr.properties
 storage.backend=hbase
 storage.hostname=zookeeper-host1,zookeeper-host2,zookeeper-host3
 storage.hbase.table=janusgraph
 storage.hbase.ext.zookeeper.znode.parent=/hbase
 storage.hbase.ext.hbase.zookeeper.property.clientPort=2181
 Just open your graph!
 graph=JanusGraphFactory.open('conf/janusgraph-hbase-solr.properties')
Optional
Optional

15. 15 #HBaseCon
with HBase
 HBase – Perfect Storage Backend for JanusGraph
Throw in an Index Backend for better performance
 conf/janusgraph-hbase-solr.properties
 index.search.backend=solr
 index.search.solr.mode=cloud
 index.search.solr.zookeeper-url=zookeeper-host1:2181/solr,zookeeper-
host2:2181/solr,zookeeper-host3:2181/solr
 index.search.solr.configset=janusgraph

16. 16 #HBaseCon
with HBase
 HBase – Perfect Storage Backend for JanusGraph
Look into more details
 Stores to Column Families
 Edge store  e
 Index store  g
 ID store  i
 Transaction log store  l
 System property store  s
 CF attributes can be set. E.g. compression, TTL.

17. 17 #HBaseCon
with HBase
 HBase – Perfect Storage Backend for JanusGraph
Look into more details
g.V().has("name", "Alice").out("knows").out("knows").values("name")
Query Plan to
Backend Store and
Index
Edge Store
Index Store
Index
provider

18. 18 #HBaseCon
with HBase
 HBase – Perfect Storage Backend for JanusGraph
Look into more details
 A store (column family) is always specified.
 Get or Multi Get
 Batch to mutate
 Key range scan
 ColumnRangeFilter
 ColumnPaginationFilter
 HBase tuning
Edge Store
Index Store

19. 19 #HBaseCon
with Google Cloud Bigtable
 Bigtable implements the HBase 1.0 client API
Need the latest version of the bigtable-hbase-1.0 artifact.
 storage.backend=hbase
 storage.hbase.ext.hbase.client.connection.impl=
com.google.cloud.bigtable.hbase1_0.BigtableConnection
 storage.hbase.ext.google.bigtable.project.id=
<Google Cloud Platform project id>
 storage.hbase.ext.google.bigtable.instance.id=<Bigtable instance id>

20. Thank you!