madtulip
Explorer
- Joined
- Mar 28, 2015
- Messages
- 64
Hello guys.
I would like to integrate my nas hardware into my matlab distributed computing network. Its a math application where i hammer for hours on GBs of data. The toolbox allows to split the iteration of i.e. independen for loops to different computers/CPUs/threads.
As i in principle like to keep the main freenas OS clean of installs i was thinking to create this working threads inside a jail. Im not sure about the performance impact of applications running in jail versus thos running in the main freeNAS OS. The application is assumed to compleetly bog down all hardware ressources and its the time gain from distributed computing im after.
So will i get big performance drops while running this 99% load processes in jails and should instead install it in the freeNAS main OS layer like it would happen with virtual machines? Or is the architekture of the software somehow different in a native freeBSD jail and its ok to run the tasks in a jail in order to have a cleaner seperation inside the system?
As the nativ jails are the same OS as the main one and i didnt see any virtualization of CPUs, drives or RAM i thought probably the jails are quite efficient in beeing executed and the main OS is able to distribute resources to the jail processes effectively without much overhead?
thanks for any advice!
I would like to integrate my nas hardware into my matlab distributed computing network. Its a math application where i hammer for hours on GBs of data. The toolbox allows to split the iteration of i.e. independen for loops to different computers/CPUs/threads.
As i in principle like to keep the main freenas OS clean of installs i was thinking to create this working threads inside a jail. Im not sure about the performance impact of applications running in jail versus thos running in the main freeNAS OS. The application is assumed to compleetly bog down all hardware ressources and its the time gain from distributed computing im after.
So will i get big performance drops while running this 99% load processes in jails and should instead install it in the freeNAS main OS layer like it would happen with virtual machines? Or is the architekture of the software somehow different in a native freeBSD jail and its ok to run the tasks in a jail in order to have a cleaner seperation inside the system?
As the nativ jails are the same OS as the main one and i didnt see any virtualization of CPUs, drives or RAM i thought probably the jails are quite efficient in beeing executed and the main OS is able to distribute resources to the jail processes effectively without much overhead?
thanks for any advice!