ポストコピー型の仮想マシン再配置機構は仮想マシンの実行ホストを素早く切り替えられるため,データセンタの運用効率を向上させるうえで有用な技術であると考えられる.しかしながら,今日一般的に利用できるまでには至っていない.先行研究におけるポストコピー型再配置機構は,既存の仮想マシンモニタ(VMM)への変更が大きく,ゲストOSの改変も必要になる点に問題がある.そこで我々は,既存VMMのへ拡張が単純でゲストOSの改変も不要な,新たなポストコピー型ライブマイグレーション機構を提案する.メモリアクセスのトラップ処理とメモリページのコピー処理をVMMの外部で実装することで,VMMへの変更量を抑えながらポストコピー型再配置を実現する.再配置性能を検証するため,SPECweb2005を用いて評価実験を行った.負荷の高いウェブサーバを実行するVMであっても,1秒以内に実行ホストを切り替えることができた.実行ホスト切替え後の性能低下は限定的であった.プレコピー型再配置に比べて,VMのすべての状態を移動する時間も短縮できた.Post-copy-based VM migration is considered a promising technology for next-generation datacenters; memory pages are transferred after a VM restarts at a destination host, thereby minimizing the time of switching the execution host. Post-copy-based migration mechanisms, however, have not yet been available in industry. Prototype implementations in prior work need major modifications to existing virtual machine monitors (VMMs), and also require special software support in guest operating systems. In this paper, we propose a simple and plain implementation of post-copy-based migration, which is implemented as a lightweight extension to KVM. It supports any guest operating systems without their modifications. The RAM of a migrated VM is mapped to a special character device, which transparently transfers memory pages on demand. Experiments were conducted by using the SPECweb2005 benchmark. A running VM with heavily-loaded web servers was successfully relocated to a destination within one second. Temporal performance degradation after relocation was alleviated by pre-caching memory pages. In addition, for memory intensive workloads, our migration mechanism moved all the states of a VM faster than a pre-copy-based migration mechanism.