Most virtualization functions are only possible through the POWER hypervisor. The hypervisor offers the possibility of dividing physical resources and making them available to the partitions. The hypervisor ensures that each partition has only access to its own resources, it isolates the individual partitions from each other. Each of the partitions can use its own instance of an operating system; the supported operating systems are AIX, IBM i and Linux, or virtual I/O servers.
Figure 1.1 shows the virtualization using a hypervisor. The hardware layer, with all physical resources such as CPUs, memory and physical slots, is below the hypervisor. Partitions do not have direct access to the hardware layer. Access is only possible via the hypervisor. The hypervisor can assign dedicated resources for exclusive use by an LPAR, e.g. dedicated CPUs or physical slots or physical memory areas, but depending on the type of resource it can also allow shared use by several LPARs, e.g. shared processors or active memory sharing. In the case of shared processors, an LPAR only gets a specified share of a processor, it can then only use the processor temporarily. The distribution to the various LPARs is carried out by the hypervisor in a time slice process.
No special processor resources are assigned to the POWER Hypervisor. LPARs communicate with the POWER hypervisor via hypervisor calls (hcalls). Under AIX, this can be monitored more closely with the lparstat command (option “-h” or “-H”).
You must be logged in to post a comment.