图数据库对比·201808

图数据库Benchmark. https://github.com/socialsensor/graphdb-benchmarks

Benchmark: PostgreSQL, MongoDB, Neo4j, OrientDB and ArangoDB. https://www.arangodb.com/2015/10/benchmark-postgresql-mongodb-arangodb/ 可以参考这篇文章看测试哪些项目。

 

 

 

Neo4j  in this article is specified to Community Neo4j.  

 

 

 

 

 

• Disadvantages
  • Data Volumn
    • The biggest disadvantage of Neo4j is the data Volumn it can supported. For Community Neo4j doesn't support Distrbution, but Enterprise one does.To achieve High Avalable goal, users need to implement distribution by their selfs.
• Advantages
  • Multiple Labels
    • An Entity supports multiple labels(types), which is very suitable for our geo data.
  • Visualization

 

任务困难

资料欠缺

 

JanusGraph 和 HugeGraph之间的比较

HugeGraph

• Domestic database with chinese documents
•  

 

JanusGraph
janusgraph-spark

      没有任何成系统的资料。
      早期Spark一共有四个模块：（1）核心（2）Sql（3）流处理Spark Stream（4）图计算模块GraphX
     I haven't found any  useful things to talk about 'Spark Janusgraph' on the top 2 pages of Google and JanusGraph official docs, except some articles about connecting the two on google.
   Though throuh my using experience on GraphX,   we can infer that  Spark can load data from JanusGraph as a Graph[Vertex,Edge] object which is a datastruct provided by GraphX.  It is also the way HugeGraph used to deal with the data of JanusGraph. Then a graph object can be parsed into VertexRDD and EdgeRDD,  Spark Sql and Spark Stream can deal with RDD. 

 

gremlin> graph = GraphFactory.open('conf/hadoop-graph/read-cassandra.properties')
gremlin> g = graph.traversal().withComputer(SparkGraphComputer)
gremlin> g.V().count()

从stackoverflow上的代码来看， 是把spark作为一个 计算器 配置到JanusGraph里， 是JanusGraph有选择的依赖Spark的关系。  

 

 

 

Comparison

Name	Trinity(Graph Engine)	Neo4j Community	OrientDB	JanusGraph	HugeGraph
Documention	very few	abundant	abundant	few
Distribution Support	yes	no		yes	yes
Visualization Tool	no	yes	yes	no	yes
Popularity Ranking	16	1	4	11
Server operating systems	Linux Windows	Linux OS X Solaris Windows	All OS with a Java JDK (>= JDK 6)	Linux OS X Unix Windows
Data scheme	yes	schema-free and schema-optional	schema-free	yes
APIs and other access methods	RESTful, HTTP API		Java API, RESTful HTTP/JSON API, Tinkerpop technology stack with Blueprints, Gremlin, Pipes	Java API, TinkerPop Blueprints, TinkerPop Frames, TinkerPop Gremlin
Supported programming languages	C# C++		All common languages	Clojure Java Python
Replication methods			Master-master replication	yes
Concurrency	yes		yes	yes

 

Trinity中的边作为Node的属性存在，本身不能具有属性；JanusGraph和Neo4j的边本身是一个对象，可以具有属性。

 

Reference

https://db-engines.com

System Properties Comparison Graph Engine vs. JanusGraph vs. Spark SQL

 

 

 

	Supporting Distributed	Need to pay for Distribution

 

 

Name JanusGraph successor of Titan  Xexclude from comparison Neo4j  Xexclude from comparison   Description A Graph DBMS optimized for distributed clusters It was forked from the latest code base of Titan in January 2017 Open source graph database   Primary database model Graph DBMS Graph DBMS   DB-Engines Ranking measures the popularity of database management systems Trend Chart Score 0.54 Rank #163   Overall   #11   Graph DBMS Score 40.92 Rank #22   Overall   #1   Graph DBMS   Website janusgraph.org neo4j.com www.sequoiadb.com Technical documentation docs.janusgraph.org/­latest neo4j.com/­docs www.sequoiadb.com/­en/­index.php?m=Files&a=index Developer Linux Foundation; originally developed as Titan by Aurelius Neo4j, Inc. Sequoiadb Ltd. Initial release 2017 2007 2013 Current release   3.3.5, April 2018   License Commercial or Open Source Open Source Apache 2.0 Open Source GPL version3, commercial licenses available Open Source Server: AGPL; Client: Apache V2 Cloud-based Only available as a cloud service no no no Implementation language Java Java, Scala C++ Server operating systems Linux OS X Unix Windows Linux Can also be used server-less as embedded Java database. OS X Solaris Windows Linux Data scheme yes schema-free and schema-optional schema-free Typing predefined data types such as float or date yes yes yes oid, date, timestamp, binary, regex XML support Some form of processing data in XML format, e.g. support for XML data structures, and/or support for XPath, XQuery or XSLT. no   no Secondary indexes yes yes pluggable indexing subsystem, by default Apache Lucene yes SQL Support of SQL no no SQL-like query language APIs and other access methods Java API TinkerPop Blueprints TinkerPop Frames TinkerPop Gremlin TinkerPop Rexster Cypher query language Java API Neo4j-OGM Object Graph Mapper RESTful HTTP API Spring Data Neo4j TinkerPop 3 proprietary protocol using JSON Supported programming languages Clojure Java Python .Net Clojure Elixir Go Groovy Haskell Java JavaScript Perl PHP Python Ruby Scala .Net C++ Java PHP Python Server-side scripts Stored procedures yes yes User defined Procedures and Functions JavaScript Triggers yes yes via event handler no Partitioning methods Methods for storing different data on different nodes yes depending on the used storage backend (e.g. Cassandra, HBase, BerkeleyDB) none Sharding Replication methods Methods for redundantly storing data on multiple nodes yes Causal Clustering using Raft protocol available in in Enterprise Version only Master-slave replication MapReduce Offers an API for user-defined Map/Reduce methods yes via Faunus, a graph analytics engine no no Consistency concepts Methods to ensure consistency in a distributed system Eventual Consistency Immediate Consistency Causal and Eventual Consistency configurable in Causal Cluster setup Immediate Consistency in stand-alone mode Eventual Consistency Foreign keys Referential integrity yes Relationships in graphs yes Relationships in graphs no Transaction concepts Support to ensure data integrity after non-atomic manipulations of data ACID ACID Document is locked during a transaction Concurrency Support for concurrent manipulation of data yes yes yes Durability Support for making data persistent yes Supports various storage backends: Cassandra, HBase, Berkeley DB, Akiban, Hazelcast yes yes In-memory capabilities Is there an option to define some or all structures to be held in-memory only.     no User concepts Access control User authentification and security via Rexster Graph Server Users, roles and permissions. Pluggable authentication with supported standards (LDAP, Active Directory, Kerberos)

 

 

https://www.sohu.com/a/199849366_505818