使用 Neo4j 处理大型数据集的经验？
有人有过使用 Neo4j 处理 TB 级数据集的经验吗？我想听听您对 Neo4j 性能的体验
Has anyone gone any experience of using Neo4j with terabyte sized datasets? I would like to hear about your expereinces with how Neo4j performs
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（3）
只要您的磁盘足够大、速度足够快，并且内存允许缓存数据的相关（热）部分，您就不会遇到问题。
有优化用于根据特定需求调整 Neo4j 数据存储。
否则，这取决于您的数据集的类型。查询性能不应该成为问题，如果您必须执行大量索引查找来连接导入的节点，则插入性能可能会受到影响（但 Neo4j 团队正在努力解决这一问题）。
也许您应该加入 Neo4j 邮件列表，以更一致地回答您的所有问题。
As long as your disk is large and fast enough and your memory allows for caching of the relevant (hot) portion of your data, you shouldn't run into issues.
There are optimizations for tuning the Neo4j datastore to specific needs.
Otherwise it depends on the kind of your dataset. Query performance shouldn't be an issue, insert performance might suffer if you have to do a lot of index lookups for joining imported nodes (But the Neo4j team works on that).
Perhaps you should join the Neo4j mailing list to answer all your questions more consistently.
我们一直在使用 Neo4j 存储用户及其关系的图，现在大小约为 10 000 个节点和 400 000 个关系，图结构中支持的某些操作（例如获取用户 Neo4j 的朋友）非常快。
它始终取决于您要在数据库上运行的查询以及存储数据库的服务器计算机。
We have been using Neo4j storing a graph of users and their relations with an approximate size now of 10 000 nodes and 400 000 relations,certain operations which are supported in a graph structure like getting friends of a user Neo4j is pretty fast.
It always depends on what queries you are going to run on the database and also the server machine storing your database.
我使用 neo4j 来处理具有 4 000 000 个节点和 42 000 000 个边的图，效果非常好。
尝试找到两个随机节点之间的最短路径，花费了不到 100 毫秒。检索邻居的邻居，包括朋友、朋友的朋友、朋友的朋友的朋友也几乎不需要时间，而同一台机器上的关系数据库允许你去吃午饭，直到它执行。
I use neo4j for handling the graph with 4 000 000 nides and 42 000 000 edges and it works great.
Have tried to find the shortest path between two random nodes and it took less than 100 ms. Retrieving the neighborhood of a neighbor, including friends, friends of friends and friends of friends of friends also takes almost no time while the relational database on same machine allows you to go for a lunch until it executes.