On Mon, 2003-03-17 at 19:13, Orlando Andico wrote:
> On 17 Mar 2003, Dean Michael Berris wrote:
> ..
> > distributed shared memory (DSM) is not what i am planning to achieve
> > here. it's really something more general -- a distributed virtual
> > machine. DSM is only one part of my problem and something that will be
> 
> why not use mosix and be done with it?
> unless of course you're doing another 'reinvent the wheel with my thesis' 
> kind of project.  :)
> 

ok, here we go. =)

i've looked at mosix, and yes it kinda does what i'm planning to do. but
the crux of my thesis (to be approved) would be in the realm of load
balancing. i'd like to create a virtual machine running on different
nodes that are part of a single system. the details would be way too
much to discuss here, but in a nutshell, i'd try nonetheless.

imagine having a group of computers in an office running linux, then
having each one run my software. as these nodes are being used by the
people, the virtual machine runs in the background taking advantage of
the idle time of the computer. to not get in the way of the each node's
performance, the virtual machine should be able to balance the load
among the nodes working for it. example, if a certain node suddenly gets
its load up (running an SDL based game on a computer without a video
card that has its own GPU) then the work that should be allocated to it 
should decrease. the work that is supposed to be allocated to it, must
be either held, or given to another node for the maintaining the
performance of the whole virtual machine.

i think this is what condor does, but im after the performance
enhancement by either developing a load balancing scheme for the said
setup, or study the existing load balancing schemes that are already out
there.

well, that was in the nutshell, and it is easier said than done. =)

> no offence meant -- my thesis was a 'reinvent the wheel' too.
> 

none taken... ;)

but yes, do you know how aluminum mags were created? hehehe... ;)

basically, its not reinventing the wheel per se -- but redoing it with a
twist. =) that i think is enough justification to say that we need
thesis that 'try-and-reinvent-the-frikin-wheel'. =)

-- 
-=[mikhail]=-

aka Dean Michael C. Berris
mobile +63 917 8901959
work +63 49 5680024
http://free.net.ph/Members/mikhailberis

_
Philippine Linux Users Group. Web site and archives at http://plug.linux.org.ph
To leave: send "unsubscribe" in the body to [EMAIL PROTECTED]

Fully Searchable Archives With Friendly Web Interface at http://marc.free.net.ph

To subscribe to the Linux Newbies' List: send "subscribe" in the body to [EMAIL 
PROTECTED]

Reply via email to