Cloud4Home— Enhancing Data Services with @Home Clouds

Motivation:

End devices are released much more frequently than server systems replacement cycle. Purely end-point based solutions cannot take advantage of the large storage and computational capacities present in large scale datacenters. The Cloud4Home project and approach described in this paper can solve this problem by exploiting the efficiencies and opportunities presented by flexible and combined @home and @datacenter operation.

Main points:

VStore++ is a set of Cloud4Home services implementing methods for data storage and manipulation that enhance what can be provided by solely @home or @datacenter service realizations.

VStore++’s implementation for a prototypical home environment, with desktop and handheld devices, uses the Xen hypervisor to attain operational independence from vendor-specific solutions.

Home security and video streaming services realized with VStore++ exhibit improved performance properties compared to prior service realization, including services realized in Amazon’s EC2 cloud.

For home device cooperation and active resource management, VStore++ uses

(i)            a dynamic overlay layer implemented with the lightweight Chimera peer-to-peer overlay system;

(ii)           a distributed key-value store for data accesses and dynamic resource monitoring.

Experimental evaluations of the VStore++ system and approach demonstrate multiple interesting facts, including

(i)                   the tradeoffs in using @home vs. @datacenter resources;

(ii)                 the advantages derived from judiciously using both;

(iii)                future work in terms of scaling Cloud4Home functionality to larger systems and to other sets of services.

Trade-offs:

The Cloud4home system may has less security than the face detection.

The Cloud4home system may waste considerable resources, including electricity and storage, if it runs all the time, since we may only use this system in short periods of time each day.