Re: openMosix vs KVM, Xen,...et al

Hi All,
	I think people may have missed the point regarding using KVM (et al.).  I 
believe the point was not to make openMosix a virtualization, but to use 
virtualization in order to implement openMosix.  
	openMosix is about moving CPU intensive programs to "real" CPUs.  openMosix 
2.4 does this by migrating processes from one computer to another. 
Alternatively, a VM would be created for each process of interest. Then one 
could migrate an entire VM, containing the CPU intensive process of interest, 
to other computers.
"Migrating a VM" http://kvm.qumranet.com/kvmwiki/Migration
	The one advantage migrating a VM would have over the 2.4 openMosix is that 
there is no "head-node" single point of failure.  Another is that one could 
snapshot the VM for recovery in case of power failure.  Shared memory (within 
the VM) and all types of threads would be automatically supported.
	The main disadvantage (at least how I currently imagine it) is that openMosix 
would be less transparent and more like XGrid or Condor.  One would have to 
package up the processes in a VM, then start the VM and let it migrate.  If 
this could be automated so that it looks like one is just executing a command 
then this would be a very nice implementation.
	Another disadvantage might be the memory footprint of the VM.  This could 
probably be small though if it were the size of Linux on embedded devices.

	To repeat, no we're not playing buzzword bingo, we're simply suggesting 
another way of distributing CPU intensive programs between real CPUs.

C.


-------------------------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc.
Still grepping through log files to find problems?  Stop.
Now Search log events and configuration files using AJAX and a browser.
Download your FREE copy of Splunk now >>  http://get.splunk.com/