I have a dataset of 10 millions in MySQL. Im using Hibernate search. My spatial query is taking around 4 sec for search in 100 km radius. Is it expected or should it be more faster? Im running my server on a m3.xlarge Amazon EC2 instance.

Hi,
normally I'd expect it to be significantly faster, but it depends on several factors.

For example, what's the density of your points? And what strategy are you using for Spatial indexing?
I can't really tell you what performance I'd expect as that's highly specific to your applidation, your data and even your data (even if I had your same source code, but had to make up different data the behaviour would be different).

Also, keep in mind that the first queries are rather slow, it takes some million queries to properly warmup both the JVM and the operating system disk caches.
We can help you figure out what could be tuned to improve performance but you'll have to run some diagnostics on the systems; for a starter, could you post some logs from "vmstat" running periodically during high load?

Also, make sure you use the latest version as there are always performance improvements being added. You might even want to try latest preview for version 5.