Power 虚拟化(高级篇)下精品.pptVIP

* From this page, the three Figures show the CPU consumption on the mobile LPAR and the Virtual I/O servers, for the three different levels of load. From these charts, it is apparent that the MSPs (run on the Virtual I/O servers) are busy during active migrations, transferring the mobile LPAR’s state from the source to the destination system. The speed at which the MSPs can transfer data is mainly limited by available CPU resources on the Virtual I/O servers and/or network bandwidth. In the test configuration, the gigabit network dedicated for the MSPs is maxed-out at about 110Mbytes/second during memory copy. * * The source and destination mover service partitions use this virtual device to communicate with the POWER Hypervisor to gain access to partition state. VASI 是 MSP 和 Hypervisor 之间进行通信的一个虚拟设备。在源系统上，MSP 通过 VASI 从 Hypervisor 获取分区的运行时状态，然后通过网络传送给目标系统的 MSP；在目标系统上，MSP 通过 VASI 把分区的运行时状态传送给 Hypervisor。在 IVM 和 HMC 两种情况下，VASI 都是必须的。在早期支持动态分区迁移的 VIOS 版本（比如 VIOS ）中，VASI 是需要用户自己去创建的；而在较新版本的 VIOS（比如 VIOS ）中，VASI 是系统自动创建的。 IVM 所在的 VIOS 的 MSP（Mover Service Partition）属性是默认打开的，没有任何选项用于打开或者取消 MSP 属性；这与 HMC 上的动态分区迁移不同，后者必须激活 VIOS 的 MSP 属性才能进行分区迁移。 VASI: virtual 异步service Interface * * * * TPC-C, an OLTP system benchmark, simulates an environment in which a population of terminal operators execute transactions against a database. The benchmark is centered on the principal activities and transactions of an order-entry environment. These transactions include entering and delivering orders, recording payments, checking the status of orders, and monitoring the level of stock at the warehouses. Note that IBM is currently the only vendor publishing TPC-C results with virtualization technology such as PowerVM. Figure 1 shows the TPC-C benchmark results at 4, 8, 16, and 64 cores. Even with virtualization technologies inherently included as discussed above, the greater than linear scaling of the results is noteworthy. * AIM7 Multiuser Benchmark, a well known open source benchmark, was