Junhao wrote:
On 04/03/2010 12:30 PM, Yong Han wrote:
Hi,
any one has any idea how can i go about experiment grid computer? the
main objective is to combine the processing power into 1 virtualised
system. is this the way?
thanks!
Can you elaborate on your intended purpose? "combine the processing
power into 1 virtualised system" is too vague...
Sounds to me like 'Single System Image':
http://en.wikipedia.org/wiki/Single-system_image
With EC2, you have 'User selectable kernels' (but only approved people
can upload them). You could potentially use a MOSIX like linux kernel on
EC2 to combine hundreds of virtual machines into a big 'Single System
Image' e.g. create a tightly coupled cluster.
http://developer.amazonwebservices.com/connect/entry.jspa?externalID=1345
Although, nowadays this architecture has become less important as this
architecture was initially needed to allow legacy multiprocess
applications to work unchanged on a distributed cluster.
Nowadays you would just design your app to partition itself into many
small independent jobs. You also get much more resilience with a more
loosely coupled cluster (i.e. grid).
_______________________________________________
LUGS Mailing list - [email protected]
List FAQ: http://wiki.lugs.org.sg/LugsMailingListFaq
Info page: http://www.lugs.org.sg/mailman/listinfo/slugnet
To unsubscribe send an empty email to: [email protected]
