50 likes | 148 Views
Explore research issues on offloading decisions in OS, considering network status, loading, and virtual core for computation-intensive and I/O-intensive tasks. Analyze conditions for remote functions, automation, and storage integration. Discover strategies for returning results to local core and balancing local and remote storage in virtual environments.
E N D
Research Issues • Offloading decision • Under what conditions should OS make offloading decisions. • Network status • Current loading • …etc. • Virtual core • computation-intensive -> I/O-intensive(wait for results)
Research Issues(Cont.) • Remote functions • Should be stateless and computation-intensive. • Modifications • By application developer. • Automation? • Not every function can be run on virtual core. • Foo(){ • intintermediateData = …; • Bar(intermediateData, …); • }
Research Issues(Cont.) • Return results to local core. • How to decide waiting for remote function or not. • Recovery.
Virtual Storage • Some applications need local files as input. • It is not practical for a remote function to read files from local storage. • Virtual storage • Combine local and remote storage. • We can upload and synchronize the files to remote storage in background. • The remote functions read the file directly from remote storage.