Question

我仍然怀疑orientdb的性能。纯粹插入图形的最高性能是2000 edge / s，而更新和插入的最高性能是250 edge / s。

我的代码在这里，与@Nandna Chowdhury非常相似

仅限插入的代码：

  def insertiongraphtest(graph: OrientGraphNoTx): Unit ={
println(new Date())
for(vcount <- 1 to 1000){
  val vertex1 = graph.addVertex("class:CLASS1",
    "prop1",vcount.asInstanceOf[java.lang.Integer],"prop2","22","prop3","3333" )
  for(vc <- 1 to 1000){
    val vertex2 = graph.addVertex("class:CLASS2",
      "prop1",(vcount+vc/1000).asInstanceOf[java.lang.Integer],"prop2","22","prop3","3333" )
    graph.addEdge(null,vertex1,vertex2,"v12v2")
  }
  graph.commit()
}
println(new Date())
}

插入和更新的代码：

  def insertupdategraphtest(graph: OrientGraphNoTx): Unit ={
println(new Date())
val vertex1type = graph.createVertexType("CLASS1")
vertex1type.setOverSize(2)
val vertex2type = graph.createVertexType("CLASS2")
vertex2type.setOverSize(2)
graph.createKeyIndex("prop1",classOf[Vertex] ,
  new Parameter("type", "UNIQUE"),
  new Parameter("class", "CLASS1"))
graph.createKeyIndex("prop1",classOf[Vertex] ,
  new Parameter("type", "UNIQUE"),
  new Parameter("class", "CLASS2"))
var vertex1:OrientVertex = null
var vertex2:OrientVertex = null
for(icount <- 1 to 100){
  for(vcount <- 1 to 10){
    val v = graph.getVertices("CLASS1.prop1",vcount).asScala.
      asInstanceOf[Iterable[OrientVertex]].iterator

    if(v.hasNext){
      while (v.hasNext){
        vertex1 = v.next()
        vertex1.detach()
        vertex1.setProperty("prop2",new Date().toString)
        vertex1.attach(graph)
        vertex1.save()
      }
    }else{
      vertex1 = graph.addVertex("class:CLASS1",
        "prop1",vcount.asInstanceOf[java.lang.Integer],"prop2",new Date().toString,"prop3","3333" )
    }

    for(vc <- 1 to 1000){
      val v2 = graph.getVertices("CLASS2.prop1",vc).asScala.
        asInstanceOf[Iterable[OrientVertex]].iterator
      if(v2.hasNext){
        while (v2.hasNext){
          vertex2 = v2.next()
          vertex2.detach()
          vertex2.setProperty("prop2",new Date().toString)
          vertex2.attach(graph)
          vertex2.save()
        }
      }else{
        vertex2 = graph.addVertex("class:CLASS2",
          "prop1",vc.asInstanceOf[java.lang.Integer],"prop2",new Date().toString,"prop3","3333" )
      }
      val edgeit = vertex1.getEdges(vertex2,Direction.OUT,"v12v2").asScala.iterator
      if(edgeit.hasNext){
        while (edgeit.hasNext){
          edgeit.next()
        }
      }else{
        graph.addEdge(null,vertex1,vertex2,"v12v2")
      }
    }
    println(icount,vcount,new Date())
  }
}
}

在我的代码中，我几乎都遵循所有调优技巧：

1.graph.declareIntent（new OIntentMassiveInsert（））

2.indexing

3.vertex2type.setOverSize（2）

4.使用plocal

我的问题在这里：如果我需要检查数据库以决定是否插入或更新，那么提高性能的最佳方法是什么？

我从笔记本电脑上得到了结果。

有人对orientdb有适用的提示吗？

谢谢！

在orientdb中插入和更新的性能？任何人都可以提供一些额外的提升技巧吗？

0 个答案: