Inter-Datacenter Migration MechanismWe study the “tracking the moon/sun” computing paradigm where VM locations are optimized around the world for better energy saving; e.g., the system automatically moves virtualized datacenters to locations with excess electricity. We are developing wide-area VM migration technologies that can efficiently transfer whole VM states to remote datacenters. This project also explores disaster recovery and mitigation techniques that can transparently evacuate IT infrastructure to safe locations in case of a severe earthquake.It is necessary to reduce migration data more drastically for relocating a bunch of VMs among datacenters. Proactive Disk Block Caching over Wide-area Network - xNBD-WANWe are extending our hypervisor-independent live storage migration mechanism to support wide-area networks more efficiently. We propose a mechanism to improve live storage migration across WAN. The key idea is to reduce the amount of data to be transferredby proactively caching virtual disk blocks to a backup site during regular VM operation. As a result of pre-cached disk blocks, the proposed mechanism can dramatically reduce the amount of data and consequently the time required to live migrate the entire VM state. Reproducing Memory Page Contents without TransferringThis is joint work with University of Tokyo. It's ongoing. See the publication list for the latest update. Taming concurrent Live Migrations over Wide-area NetworksThis is joint work with ACIS Lab of University of Florida. It's ongoing. See the publication list for the latest update. Transparent IP Packet Forwarding with Mobile IPv6 InfrastructureKagemusha is a transparent Client Mobile IPv6 Tunneling mechanism that allows light-weight overlay networks among geographically-distributed cloud systems. No MIPv6 program needs be installed into a guest operating system. Kagemusha is fully compatible with existing home agents (HAs) in the MIPv6 infrastructure. See the dedicated project page.Selected Publications
Disaster Mitigation and RecoveryThis project studies the effectiveness of movable virtualized datacenters in keeping IT services alive during and after a disaster by investigating the joint usage of VM migration (live or using checkpoints), virtual networking, and shared/replicated storage for VM images.
Multi-site Virtual ClusterA multi-site virtual cluster
is a novel concept for distributed resource collabolation among
datacenters. It allows flexible and scalable resource management far
beyond limitations at a single site. It is an enabling technology of
Cloud Federation among distributed datacenters and organizations. ![]() Publications
|