[GraphHopper] Miltiple GraphHopper instances for best productivity

[Alexander Cherepanov]

Hello.
We are trying to start multiple threads with GraphHopper instances and send them calculate orders with pooling engine.
All GraphHopper works with option 
graph.dataaccess=MMAP_STORE_SYNC
With JRE7 64bit on Windows 7 Professional our application works well.

But there is a problem on CentOS 6 with OpenJDK:
OpenJDK Runtime Environment (rhel-2.5.1.2.el6_5-x86_64 u65-b17)
OpenJDK 64-Bit Server VM (build 24.65-b04, mixed mode)

on command TOP columns RES and SHR only incrementing on the time, and if we start much more threads - then much more resources are used.
Why this happens?


--