问题
I have read this article. It states, that Neo4j can scale horizontally, but only to increase read performance and fault tolerance... so the stored graph is copied to each server in a cluster. But what if I have a dataset that is larger than one server can store? Does Neo4j fail in this situation? Do I have to scale vertically in this situation and buy larger HDD?
Thank you
回答1:
Yes. You need enough hard drive space to contain the full graph on all nodes of the cluster, no way around that.
If you're instead referring to RAM instead of hard drive space, then it isn't necessary to have all of the db in memory (defined by the pagecache setting in neo4j.conf), but that will mean you'll hit the disk on all pagecache misses.
Here's the memory configuration section in the docs for details.
来源：https://stackoverflow.com/questions/50200734/neo4j-scalability