Unit 5 Lab 5.5

VSphere 5.5 has introduced vNUMA which allows the Guest OS to be aware that they are running on a NUMA (Non Uniform Memory Architecture) topology. In the past and currently the big issue was that EIT could not build virtual machines with CPUs that would span over multiple physical sockets as this would create performance issues and contention when requesting resources. When this occurs it will use its own physical socket memory as well as the socket it has being spanned across causing that its own NUMA-aware resource management features would become problematic causing performance issues. EIT had to make sure that vCPU did not exceed the physical socket and that it is kept in its own NUMA node. The problem that arose was to not build big virtual …show more content…
The problem with this type of deployment is that it becomes difficult for storage administrators to investigate effectively issues that are experienced from a datastore level making troubleshooting more complex. A LUN (logical unit number) consists of a chunk of storage given to the hypervisor and presented as a datastore in virtual center .The other problem is that you cannot utilize the datastores fully as you need to have enough capacity for virtual machine snapshots and the growth of virtual machine files. If a few datastores become problematic it can affect that hypervisor and cause it to become unresponsive. When you delete a virtual machine the array itself cannot reclaim the storage blocks from the hypervisor. Although the datastores are assigned as thin provisioned disks to the hypervisor the storage admins still cannot control individual LUNs from the array side and in addition would become the virtualization administrator’s problem. Snapshots can cause the datastore to run out of capacity, which causes that virtual machines to enter into a suspended state if there is not enough capacity