Performance impact on number of nodes

[AJ]

Hi,

I am using Neo4j for my travel database where every location is a node and hotels, restaurants are child nodes of location nodes. Right now number of nodes is less but later own it will easily cross a hundred thousand nodes. So my question is

1) Has any body been using Neo4j with more than a hundred thousand nodes in production environment, if yes what kind of optimizations from modelling perspective they had to do?

2) If number of nodes is an important factor in performance and scalability than should we model such as to keep the nodes to minimum. 

Thanks in advance!

-

Ankit

[Wes Freeman]

Based on my limited experience so far, it seems like the biggest things are:

- having enough RAM (where enough depends largely on the kinds of things you are doing)
- keeping your queries local and limit the start points to a reasonable number (don't do start n=node(*) match n-[r*]-m return n, m; and expect it to finish in this century with a big graph)

Wes

--