Neo4j 中单个条目的每个节点和每个边的开销是多少
我有一个整数图(比如 12 字节)。 每个节点有 200 个出边。 我有 100*10^6 个节点。
给定“关键”权重 12b,当它在 Neo4j 中表示时,总大小是多少? 边缘成本是多少?
这么大的neo4j可以查询吗? --- 我的意思当然是一些内部约束,查询的时间将表达 I/O 性能(我猜?)。
我对查询感兴趣:哪个节点连接到节点 X。JVM
限制怎么样?有机会碰上什么东西吗?
I have a graph of integers (say 12 bytes).
Each node has 200 outgoing edges.
I have 100*10^6 nodes.
Given the 'key' weights 12b what is the total size, when it is represented in neo4j?
Whats is the cost of edge?
It this size of neo4j possible to query? --- I mean of course some internal constrains, the time of the query would express I/O performance (I guess?).
I am interested in query: which node is connected to node X.
What about JVM limits? Any chance to bump into something?
如果你对这篇内容有疑问,欢迎到本站社区发帖提问 参与讨论,获取更多帮助,或者扫码二维码加入 Web 技术交流群。
绑定邮箱获取回复消息
由于您还没有绑定你的真实邮箱,如果其他用户或者作者回复了您的评论,将不能在第一时间通知您!
发布评论
评论(1)
当前的大小限制为 320 亿个节点、320 亿条关系和 640 亿条财产。一般来说,查询时间是关于需要触及多少图,图的大小关系不大。根据您的查询,您可能较少使用缓存,这完全取决于 I/O 性能。关于 Neo4j 中的缓存 的文档应该会给您一些有关磁盘/内存的提示成本。
The current limits on size is 32 billion nodes, 32 billion relationships and 64 billion properties. In general query time is about how much of the graph needs to be touched, the size of the graph matters very little. Depending on your queries, you may have less use of the caching and then it's all down to I/O performance. The documentation on caches in Neo4j should give you some hints on disk/memory costs.