[工学]支撑云计算的虚拟化技术.ppt

集中的计算以服务的形式提供给大众，那么提供什么呢？ 可以提供各种软件（SaaS）、可以提供整套应用（PaaS）、也可以提供（虚拟机的）计算机（系统）（IaaS）； 还可以提供信息存储服务（Storage as a Service） * * * 核心问题是: 当虚拟机对诸如高速缓存、内存、I/O带宽等计算资源的需求发生变化时，如何能察觉或预测到，并能及时做出重新优化的反应。 * * * 预测精度 Monotonic (40 ~ 170 MB)? Random (40 ~ 170 MB)? * * 数据中心全局内存调控 本地局部调整 远程内存 在线迁移 * * Ballooning Works as a driver or kernel module in the guest OS From the guest OS’s view: Just allocate or release memory pages To decrease memory allocation: Apply for memory pages from guest OS Those pages are then returned to hypervisor To increase memory allocation: Free those allocated pages The amount to increase/decrease is typically specified manually * * 本地调整：气球技术（ballooning） * * REMOCA: Hypervisor Remote Disk Cache Motivation To improve paging performance for memory-intensive or I/O-intensive workloads by utilizing free memory resource on another physical machine Solution The remote memory plays the role of a storage cache between a VM’s virtual memory and its virtual disk devices. In most cases, the network latency is much lower than the disk latency (1~2 magnitude) * * Network latency is lower The round-trip latency of requesting a block of continuous data from the disk (disk reads) and over 1 Gbps ethernet using TCP connections. * * REMOCA: The design of REMOCA Local Module: a ghost buffer REMOCA is an exclusive cache Remote Module: the memory service * * The Remote Paging Model Remote memory serves a storage cache for the VM Lies between VM’s mem and its virtual disk VMM intercepts all disk accesses from a VM VMM puts a disk block to the cache when necessary If the access hits the remote cache, the request is satisfied without real disk I/O Remote cache is managed locally in VMM Block is indexed by disk addresses (dev, sec) Low consulting overhead Transparent to the guest OS Easy to enable/disable * * Summary of REMOCA REMOCA can efficiently alleviate the impact of thrashing behavior, and also significantly improve the performance for real-world I/O intensive appli