Dynamic and Power Optimization of VMM 2012-Hyper-V + XenServer + VMWare
In the previous post about VMM 2012 I covered the ability to use the 3 migration of VMs technology between host XenServer, VMWare and Hyper-V. Everyone can be in the same group and using the PRO Tips. Details in http://bit.ly/ndwjA0
But now let’s talk about two new features and how they work:
- Dynamic Optimization – Manages the aggressiveness with which VMs are moved across the nodes in "hot"
- Power Optimization – turns off and power on cluster nodes as the use of resources
This feature will manage with what level of aggression will do load balancing on the hosts. Is compatibile with VMWare and XenServer since the BMC is installed on the hosts. Note however that the process of migration of VMs between hosts occurs the same OS.
Note on the screen below that it is possible to manually set the frequency at which this process will run. Times moves caused very high excessive VMs between hosts, long times can generate slow on a host until the VMs to be moved. The ideal is 10 to 30 minutes for the detection and troubleshooting.
Below we see the configuration considered ideal for VMM detects moves need of VMs. In the example we have 30% of CPU, 512 MB of memory free and do not take into account the IOPS and Network because these two common items are shared among the nodes of a cluster and are not moves optimized with between nodes.
NOTE: Recalling once again that this configuration is done in groups that can contains Hyper-V, VMWare and Xen Server and that the moves will happen between these servers with the same OS and not between different OSes. Besides it is necessary in the case of VMWare and Xen Server that are in the cluster.
This new feature is very interesting, taking into account that many cluster has twice the average need taking into account the peaks. The Dynamic Optimization helps when the peak occurs to distribute the VMs, but and when there are plenty of resources?
Power Optimization will shut down the nodes which are not necessary when the use of the hosts have nodes reducing umca given capacity and at the chosen time.
In the example below we will shut down the host provided that the use of other nodes moves the VM are not above 40% and 1 GB of RAM, and provided that it is in night time or weekend.
The shutdown process shutdown is a being which the resetting is accomplished by packages Wake On Lan (WOL) that need to be enabled in the BIOS of the host. Additionally in ESX hosts and Xen Server is necessary to take the BMC, as well as in Dynamic Optimization.
Besides, there is a proportion for this feature:
- Cluster of 4 to 5 nodes – 1 node is turned off
- Cluster of 6 to 8 nodes – 2 we will shut down
- Cluster of 9 to 10 nodes – 3 we will shut down
- Above of 10 – 1 additional node can be switched off each 2
NOTE:Power Optimization feature only work between cluster nodes and does not host-to-host.