JanusGraph·上手JanusGraph

目录

图存储比较

社区

Install and Start gremlin.sh

图存储比较

• titan 停止更新, janus 还未发布。
• neo4j 单机性能超高，分布式瓶颈大。

中文入门资料

图数据库JanusGraph介绍及使用(一)：简介 https://blog.csdn.net/gobitan/article/details/80939224

图数据库JanusGraph介绍及使用(二)：架构 https://blog.csdn.net/gobitan/article/details/80939276

图数据库JanusGraph介绍及使用(三)：安装与初步使用：https://blog.csdn.net/gobitan/article/details/81068459

JanusGraph的schema及数据建模

JanusGraph查询和数据类型. https://docs.janusgraph.org/latest/search-predicates.html

社区

• https://groups.google.com/forum/#!forum/janusgraph-users Goole论坛
• Chat: join us on Gitter
• Stack Overflow: see the janusgraph tag
• Twitter: follow @JanusGraph for news and updates
• Mailing lists:
  • janusgraph-users (at) googlegroups.com (archives) for questions about using JanusGraph, installation, configuration, integrations To join with a Google account, use the web UI; to subscribe/unsubscribe with an arbitrary email address, send an email to:
    • janusgraph-users+subscribe (at) googlegroups.com
    • janusgraph-users+unsubscribe (at) googlegroups.com
  • janusgraph-dev (at) googlegroups.com (archives) for internal implementation of JanusGraph itself To join with a Google account, use the web UI; to subscribe/unsubscribe with an arbitrary email address, send an email to:
    • janusgraph-dev+subscribe (at) googlegroups.com
    • janusgraph-dev+unsubscribe (at) googlegroups.com

Architecture

一般来说，应用程序可以通过如下两种方式与JanusGraph交互：

• 嵌入式JanusGraph：它与执行Gremlin查询语言的应用程序运行在同一个JVM中。查询执行，JanusGraph图缓存和事务处理都发生在同一个JVM中，但后端的数据存储可以是本地也可以在远程。
• JanusGraph服务器：通过提交Gremlin语言到JanusGraph服务器来交互。

下面是JanusGraph的架构图

Gremlin是Apache TinkerPop的一个模块。

实战笔记

• JanusGraph的EdgeLabel和PropertyKey的name不能相同。

About JanusGraph as RDF Store or Sparql Supporting

• Simple Conclusion
  • JanusGraph provides native support for the property graph data model exposed by Apache TinkerPop and uses Gremlin as its query language. It does not have native support for RDF or SPARQL. That being said, you could write custom scripts to ingest RDF and transform it into a property graph model. Daniel Kuppitz started work on transforming SPARQL queries into Gremlin, and that effort continues on at https://github.com/LITMUS-Benchmark-Suite/sparql-to-gremlin （32 stars，4 forks, It is based on the Apache Jena SPARQL processor ARQ）
  • 来源：https://groups.google.com/d/msg/janusgraph-users/I4rW_KhcgFE/WTxIx2ocCwAJ. 2017年3月
• load rdf file to GREMLIN
  • https://groups.google.com/forum/#!topic/gremlin-users/nIE6uaSck8g
  • import org.openrdf.sail.memory.MemoryStore; g = new SailGraph(new MemoryStore()) g.loadRDF('./test2.rdf', 'rdf-xml'))
  • Invalid import definition: 'org.openrdf.sail.memory.MemoryStore'; reason: startup failed: script1532930094647133709938.groovy: 1: unable to resolve class org.openrdf.sail.memory.MemoryStore @ line 1, column 1. import org.openrdf.sail.memory.MemoryStore; ^ 1 error
  • I cant understand SailGraph well, it need me to read the Tinkpop Documention
• File Format SupportedJanusGraph supports 3 file formats that are provided via Apache TinkerPop -- Gryo, GraphML, and GraphSON.
  • http://tinkerpop.apache.org/docs/current/reference/#_gremlin_i_o You can load it like this: gremlin> graph = JanusGraphFactory.open("conf/janusgraph-hbase.properties") ==>standardjanusgraph[hbase:[127.0.0.1]] gremlin> graph.io(gryo()).readGraph("data/tinkerpop-modern.kryo") gremlin> graph.io(graphml()).readGraph("data/tinkerpop-modern.xml") gremlin> graph.io(graphson()).readGraph("data/tinkerpop-modern.json") If you have some other format file, you'll need to write code to read it in the data file, and then construct the graph elements based on the data.

Install and Start gremlin.sh

• 解压JanusGraph 0.1.1
• 安装hadoop和elasticsearch.
• 在下面这张众神图上做练习

• JanusGraph 0.1.1无需改动即可顺利运行以下命令。
  • #Start gremlin.sh
  • sudo ./gremlin.sh
  • graph = JanusGraphFactory.build().set('storage.backend', 'inmemory').set('index.search.backend', 'elasticsearch').open()
  • graph = JanusGraphFactory.build().set('storage.backend', 'inmemory').open() #不使用索引
  • GraphOfTheGodsFactory.load(graph)
    • Unknown external index backend: search Type ':help' or ':h' for help.
    • 貌似是必须要有set('index.search.backend', 'elasticsearch')，否则打不开。
  • g = graph.traversal()
  • saturn = g.V().has('name', 'saturn').next()
• 写入数据
• 同一个顶点同样的属性key写多次测试

JanusGraph·Java写数据. http://www.k6k4.com/chapter/show/aafiizxav1531746415578

graph = JanusGraphFactory.build().set('storage.backend', 'inmemory').open()

Vertex v1 = graph.addVertex("USER")

v1.property("uid", "100")

v1.property("uid").value()

==>100 v1.property("uid", "1001") #属性的修改

v1.property("uid").value()

==>1001

mgmt = graph.openManagement()

//创建了一个名字为name的属性，并设置值类型为String，且可以保存可以重复的多个值

nameKey = mgmt.makePropertyKey('name').dataType(String.class).cardinality(Cardinality.LIST).make()

mgmt.commit()

//存储并查询values为list类型的键值对

gremlin> v1.property("name","name1") ==>vp[name->name1] gremlin> v1.property("name","name2") ==>vp[name->name2]

gremlin> v1.properties("name") ==>vp[name->name1] ==>vp[name->name2]

v1.property("age", 23);

Vertex v2 = graph.addVertex("PHONE");

v2.property("phone", "13811111111");

//创建边

Edge e12 = v1.addEdge("USER_PHONE", v2);

e12.property("create_time", "2018-08-08");

graph.tx().commit();

graph.close();

Load/Import Data

• Ways
  • https://docs.janusgraph.org/latest/hadoop-tp3.html
  • gremlin-importer
  • search 'titan bulk load csv' /spark bulkload
• Installation Guide

https://github.com/vsantosu/gremlin-importer

• CSV Import guide
• https://github.com/vsantosu/gremlin-importer/wiki/CSV-import-guide
  • https://github.com/vsantosu/gremlin-importer/wiki/CSV-import-guide#csv-format-information-and-limitations
  • Always quote everything in the CSV(I know, I did not in my examples) but you should.
• node.csv
  • node.csv example：

category, id, name, born_place, salary, siblings, rank, first_battle label, numeric, string, string, numeric, numeric, numeric, date hero, 1,3-D Man, Dmitriyevka,8.46,1,7,12/17/1995 villain, 2,A-Bomb (HAS), Roma,7.6,5,5,4/3/02004

• To our dataset, the first three lines will be:

category, id, srcUri

label, numeric, string

source,1,<http://dbpedia.org/ontology/description>

• edge.csv
  • Each edge is represented as a combination of 3 rows. Source row, target row, and edge row. For example:

id, numeric, 1428 id, numeric, 1 out,worked,hours, numeric, 1,date, date, 8/1/1984

• a ***label*** with value "worked"
• a numeric field named ***hours*** with value 1, and a date field named ***date*** with value "8/1/1984"