自我网络权重计算的性能问题答案

【问题标题】：Performance issue of calculation for the weight for ego net自我网络权重计算的性能问题
【发布时间】：2018-01-16 02:50:21
【问题描述】：

我们有以下图表数据。只有 5 个顶点，但我们有很多边。如何处理以下查询？我想获取从一个节点到另一个节点的路径。我只想得到自行车路径并计算自我的重量总和。循环将是从一个到另一个，从一个到它自己。我想过滤权重计算的总和。

suresh       = graph.addVertex(label,'person','uuid','7bff1bc0-cef1-1033-8f28-d99da6cfd8a9')
robin_niu    = graph.addVertex(label,'person','uuid','e3348740-d37f-1031-8b5b-89fbb6fdad64')
hujunjie     = graph.addVertex(label,'person','uuid','5e5139c0-dbe7-102e-8780-bedba724cbf7')
clintpollitt = graph.addVertex(label,'person','uuid','d92c6340-f98b-1035-85d7-bee5d5cc5ebe')
yanjuqi      = graph.addVertex(label,'person','uuid','2d9fba40-74c7-1033-8e84-d3a6c90ad2e9')

suresh.addEdge('Communication',robin_niu,'date','2017-11-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',robin_niu,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',robin_niu,'date','2017-11-03T00:00:00Z','weight',1)
suresh.addEdge('Communication',robin_niu,'date','2017-11-04T00:00:00Z','weight',1)
suresh.addEdge('Communication',robin_niu,'date','2017-11-05T00:00:00Z','weight',1)
suresh.addEdge('Communication',hujunjie,'date','2017-12-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',hujunjie,'date','2017-12-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',hujunjie,'date','2017-12-03T00:00:00Z','weight',1)
suresh.addEdge('Communication',hujunjie,'date','2017-12-04T00:00:00Z','weight',1)
suresh.addEdge('Communication',hujunjie,'date','2017-12-05T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-10-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-10-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-10-03T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-10-04T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-10-05T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-01-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-01-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-01-03T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-01-04T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-01-05T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-11-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-11-03T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-11-04T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-11-05T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-12-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-12-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-12-03T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-12-04T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-12-05T00:00:00Z','weight',1)
suresh.addEdge('Communication',yanjuqi,'date','2017-12-05T00:00:00Z','weight',1)
suresh.addEdge('ProfilesReportingToChain',clintpollitt,'date','2016-12-01T00:00:00Z','weight',5)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-01T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-02T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
suresh.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)

clintpollitt.addEdge('Communication',robin_niu,'date','2017-12-01T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',robin_niu,'date','2017-12-02T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',robin_niu,'date','2017-12-03T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',hujunjie,'date','2017-11-01T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',hujunjie,'date','2017-12-02T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',hujunjie,'date','2017-10-03T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',hujunjie,'date','2017-12-04T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',hujunjie,'date','2017-12-05T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',yanjuqi,'date','2017-10-01T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',yanjuqi,'date','2017-10-02T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',yanjuqi,'date','2017-10-03T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',yanjuqi,'date','2017-10-04T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',yanjuqi,'date','2017-10-05T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',yanjuqi,'date','2017-01-01T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',yanjuqi,'date','2017-01-02T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',yanjuqi,'date','2017-01-03T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',yanjuqi,'date','2017-01-04T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',yanjuqi,'date','2017-01-05T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',yanjuqi,'date','2017-11-01T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',yanjuqi,'date','2017-11-02T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',yanjuqi,'date','2017-11-03T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',yanjuqi,'date','2017-11-04T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',yanjuqi,'date','2017-11-05T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-01T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-01T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-01T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-01T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-01T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-02T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-02T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-02T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-02T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-02T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-02T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-02T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-02T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-02T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-02T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-02T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-02T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-11T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-11T00:00:00Z','weight',1)
clintpollitt.addEdge('Communication',suresh,'date','2017-11-11T00:00:00Z','weight',1)

yanjuqi.addEdge('Communication',robin_niu,'date','2017-12-11T00:00:00Z','weight',1)
yanjuqi.addEdge('Communication',robin_niu,'date','2017-12-12T00:00:00Z','weight',1)
yanjuqi.addEdge('Communication',robin_niu,'date','2017-12-13T00:00:00Z','weight',1)
yanjuqi.addEdge('Communication',hujunjie,'date','2017-11-11T00:00:00Z','weight',1)
yanjuqi.addEdge('Communication',hujunjie,'date','2017-12-12T00:00:00Z','weight',1)
yanjuqi.addEdge('Communication',hujunjie,'date','2017-10-13T00:00:00Z','weight',1)
yanjuqi.addEdge('Communication',hujunjie,'date','2017-12-14T00:00:00Z','weight',1)
yanjuqi.addEdge('Communication',hujunjie,'date','2017-12-15T00:00:00Z','weight',1)
yanjuqi.addEdge('Communication',clintpollitt,'date','2017-10-01T00:00:00Z','weight',1)
yanjuqi.addEdge('Communication',clintpollitt,'date','2017-10-02T00:00:00Z','weight',1)
yanjuqi.addEdge('Communication',clintpollitt,'date','2017-10-03T00:00:00Z','weight',1)
yanjuqi.addEdge('Communication',clintpollitt,'date','2017-10-04T00:00:00Z','weight',1)
yanjuqi.addEdge('Communication',suresh,'date','2017-10-05T00:00:00Z','weight',1)
yanjuqi.addEdge('Communication',suresh,'date','2017-01-01T00:00:00Z','weight',1)

robin_niu.addEdge('Communication',hujunjie,'date','2017-12-11T00:00:00Z','weight',1)
robin_niu.addEdge('Communication',hujunjie,'date','2017-12-12T00:00:00Z','weight',1)
robin_niu.addEdge('Communication',hujunjie,'date','2017-12-13T00:00:00Z','weight',1)
robin_niu.addEdge('Communication',hujunjie,'date','2017-11-11T00:00:00Z','weight',1)
robin_niu.addEdge('Communication',hujunjie,'date','2017-12-12T00:00:00Z','weight',1)
robin_niu.addEdge('Communication',hujunjie,'date','2017-10-13T00:00:00Z','weight',1)
robin_niu.addEdge('Communication',yanjuqi,'date','2017-12-14T00:00:00Z','weight',1)
robin_niu.addEdge('Communication',yanjuqi,'date','2017-12-15T00:00:00Z','weight',1)
robin_niu.addEdge('Communication',clintpollitt,'date','2017-10-01T00:00:00Z','weight',1)
robin_niu.addEdge('Communication',suresh,'date','2017-11-01T00:00:00Z','weight',1)
robin_niu.addEdge('Communication',suresh,'date','2017-01-01T00:00:00Z','weight',1)
robin_niu.addEdge('ProfilesReportingToChain',yanjuqi,'date','2017-10-02T00:00:00Z','weight',5)
robin_niu.addEdge('ProfilesColleague',hujunjie,'date','2017-10-03T00:00:00Z','weight',2)

hujunjie.addEdge('Communication',robin_niu,'date','2017-12-11T00:00:00Z','weight',1)
hujunjie.addEdge('Communication',robin_niu,'date','2017-12-12T00:00:00Z','weight',1)
hujunjie.addEdge('Communication',clintpollitt,'date','2017-12-13T00:00:00Z','weight',1)
hujunjie.addEdge('Communication',clintpollitt,'date','2017-11-11T00:00:00Z','weight',1)
hujunjie.addEdge('Communication',clintpollitt,'date','2017-12-12T00:00:00Z','weight',1)
hujunjie.addEdge('Communication',suresh,'date','2017-10-13T00:00:00Z','weight',1)
hujunjie.addEdge('Communication',suresh,'date','2017-11-01T00:00:00Z','weight',1)
hujunjie.addEdge('Communication',suresh,'date','2017-01-01T00:00:00Z','weight',1)
hujunjie.addEdge('Communication',yanjuqi,'date','2017-12-14T00:00:00Z','weight',1)
hujunjie.addEdge('Communication',yanjuqi,'date','2017-12-15T00:00:00Z','weight',1)
hujunjie.addEdge('Communication',yanjuqi,'date','2017-10-01T00:00:00Z','weight',1)
hujunjie.addEdge('ProfilesReportingToChain',yanjuqi,'date','2016-10-02T00:00:00Z','weight',5)
hujunjie.addEdge('ProfilesColleague',robin_niu,'date','2017-10-03T00:00:00Z','weight',2)

下面的查询不能很好地工作。我该如何解决这个问题？

g.V().has('uuid','e3348740-d37f-1031-8b5b-89fbb6fdad64').as('v').
  repeat(outE().as('e').inV().as('v')).
    until(has('uuid','2d9fba40-74c7-1033-8e84-d3a6c90ad2e9')).
  store('a').
    by('uuid').
  store('a').
    by(select(all, 'v').unfold().values('uuid').fold()).
  store('a').
    by(select(all, 'e').unfold().
       store('x').
         by(union(values('weight'),
                  select('x').count(local)).fold()).
       cap('x').
       store('a').
         by(unfold().limit(local, 1).fold()).unfold().
       sack(assign).
         by(constant(1d)).
       sack(div).
         by(union(constant(1d),
                  tail(local, 1)).sum()).
       sack(mult).
         by(limit(local, 1)).
       sack().sum()).
  cap('a')

【问题讨论】：

我知道您在底部发布的查询非常好-您说它效果不佳，但不要说原因。它与性能有关吗？
对不起-您的问题标题是“性能”，所以我想这就是问题所在。您发现这种遍历在您的环境中有多慢？
实际上，当我们执行这个脚本时，它已经被我们的环境钩住了。我已经为 uuid 创建了索引。还有其他的cmets/建议吗？
此脚本无法运行。我们试图找到另一种方法来解决此问题。
但现在我们没有找到解决这个棘手问题的好方法。

标签： graph gremlin tinkerpop3 janusgraph

【解决方案1】：

此查询将永远运行，因为您的图表包含循环路径，您的查询不会考虑这些路径。我假设，您只关心最短路径（否则填充 a 集合 IMO 没有多大意义）。一个简单的更改将修复您的查询：

gremlin> g.V().has('uuid','e3348740-d37f-1031-8b5b-89fbb6fdad64').as('v').
......1>   repeat(outE().as('e').inV().as('v')).
......2>     until(has('uuid','2d9fba40-74c7-1033-8e84-d3a6c90ad2e9').
......3>             or().cyclicPath()).
......4>   has('uuid','2d9fba40-74c7-1033-8e84-d3a6c90ad2e9').limit(1).
......5>   store('a').
......6>     by('uuid').
......7>   store('a').
......8>     by(select(all, 'v').unfold().values('uuid').fold()).
......9>   store('a').
.....10>     by(select(all, 'e').unfold().
.....11>        store('x').
.....12>          by(union(values('weight'),
.....13>                   select('x').count(local)).fold()).
.....14>        cap('x').
.....15>        store('a').
.....16>          by(unfold().limit(local, 1).fold()).unfold().
.....17>        sack(assign).
.....18>          by(constant(1d)).
.....19>        sack(div).
.....20>          by(union(constant(1d),
.....21>                   tail(local, 1)).sum()).
.....22>        sack(mult).
.....23>          by(limit(local, 1)).
.....24>        sack().sum()).
.....25>   cap('a')
==>[2d9fba40-74c7-1033-8e84-d3a6c90ad2e9,[e3348740-d37f-1031-8b5b-89fbb6fdad64,2d9fba40-74c7-1033-8e84-d3a6c90ad2e9],[5],5.0]

【讨论】：