Scalable search problem

I'm trying to solve a problem. I want to create a really scalable architecture that would be a base for a website. I don't know much about searching algorithms and scalability, I tried googling开发者_JS百科 but didn't get any good results so I'd like to present my problem and if anyone can point me in the right direction, that would be great.

This is the setup...

I'll have nodes (processes running on different computers). Each node will be responsible for storing data for a number of people. For each person I'll have a fixed set of attributes. Lets presume that the attributes are going to change very often.

How can I make a scalable, fast search across all the nodes from a single node?

Starting from question itself:

"How can I make a scalable, fast search across all the nodes from a single node?"

You question shows that this "single node" is going to be bottle-neck. So if you want to make it really scalable and ask us, about direction, I would suggest you to change design on beginning into

Search across nodes, with search queries coming from many nodes

To sum it up:

You asked how to make scalable architecture one-to-many I, suggest you to think how to make it many-to-many. Than you might find many ways to look for solution depending on details. Starting from map-reduce paradigm, finishing at concrete implementations like Sector/Sphere.