Introduces post-copy VM migration where destination pulls memory pages on-demand after migration. Compares with pre-copy migration. Covers demand paging, active push, prepaging, dynamic self-ballooning, and evaluation showing reduced downtime and total migration time.
Describes VMware ESX Server memory management supporting overcommitment. Covers memory virtualization with PPN-to-MPN mapping, ballooning technique for reclaiming memory, demand paging with swap, content-based page sharing to eliminate redundancy, and proportional-share allocation.
Presents Xen paravirtualized VMM for x86 allowing multiple commodity OSes to share hardware. Covers paravirtualization approach, memory virtualization with shadow page tables, CPU virtualization using privilege rings, device I/O using shared rings, and performance evaluation.
Presents SnowFlock for sub-second VM cloning in cloud. Covers VM fork abstraction for stateful worker creation, lazy state replication using VM descriptors, memory-on-demand via copy-on-access, avoidance heuristics, and multicast distribution for parallel computing.
Explores evolution of virtual machine monitor technology from 1960s mainframes to modern cloud computing. Covers VMM implementation including paravirtualization, full virtualization, binary translation, CPU and memory virtualization, and I/O device virtualization.