DRS is not just about keeping VMs happy. DRS has many other operational benefits that keep your admins happy too. Let’s dive into the top ways that DRS makes the day in the life so much better for the vSphere admin.
Host Maintenance Mode
Maintenance windows are a part of life. vMotion has made this task so much more manageable since it allows non-disruptive changes, but administrators still have the tedious task of evacuating the VMs from a host before maintenance activities can be started. In a very dense environment this could be quite time consuming. Without DRS the administrators are making their own decisions on where to place the VMs on the hosts that will remain operational – for every single host that’s being evacuated.
With DRS, all of this is handled for the administrator. DRS will decide the best place to move each VM by looking at several different performance metrics across both the hosts and the VMs. Each VM is moved to a host that’s best for it as well as least impactful to the whole cluster – automatically.
vSphere Update Manager – Rolling Upgrades
vSphere Update Manager (VUM) is hands-down the best way to upgrade vSphere hosts. It will download the appropriate updates and allow administrators to create compliance policies to ensure that all the hosts are running the approved version of ESXi. DRS makes this a very simple process, especially with tens or hundreds of hosts. Not only can VUM update a single host, but VUM can update entire clusters as well. When a cluster is selected for an update, VUM will automatically place hosts into maintenance mode. DRS will then automatically evacuate the VMs to the remaining hosts. Once the host is in maintenance mode, VUM will apply the updates and then exit maintenance mode. VUM will then place the next host into maintenance mode. This rolling upgrade approach allows the administrator to update an entire cluster of vSphere hosts with a few simple clicks. And because DRS is making those placement decisions, you can rest assured that the VMs will be running in the best possible location.
vSphere HA Resource Rebalance
vSphere HA (HA) is such a no-brainer. Clicking just a checkbox, VMs on a failed host will be automatically restarted on healthy hosts. However, HA’s primary intention is to get VMs powered back on, not to ensure they are restarted on any particular host. When DRS is used on the cluster, vSphere HA can reach out to vCenter and request that DRS makes a placement decision where the VM should be restarted. This is great for administrators because the angry phone calls from applications owners about poor performance after a failover event are avoided. DRS will ensure a single host does not get overloaded, and will rebalance the resources in a cluster if it finds there are better hosts to run the VMs.
DRS Groups and Affinity Rules
There are several reasons a VM may need to run on a certain host, or a group of hosts within a cluster. Licensing is a prime example for this. There are some applications that are licensed per physical CPU, and it wouldn’t make financial sense to license every CPU in the cluster just for a single VM. DRS groups allow administrators to isolate where a VM can run without creating dedicated clusters or datastores. The fewer compute islands that exist in an environment, the more that overall utilization will be improved. DRS Groups allow administrations to create should and must rules. Should rules are just that, they instruct DRS that VMs should run on a host, but if there is a reason it cannot, it will allow the VM to run on another host. A must rule is absolute. It instructs DRS to never allow a VM to run outside the hosts defined in the rule.
DRS also features affinity and anti-affinity rules. These have been around since the initial version of DRS. These allow administrators to create simple rules that either run a group of VMs on the same hosts or to separate them across hosts. This ensures that app/db servers will run on the same host for maximum performance, or that front end web servers run on separate hosts.
Resource Pools
DRS also opens up the functionality of Resource Pools, which enables the vSphere cluster resources to be abstracted from the larger pool of capacity and compartmentalize the subset of resources. The subset of resources can then be assigned to different applications or business units that consume the capacity. This allows vSphere administrators to provide compute resources to a group of virtual machines and define minimum and maximum boundaries without creating separate physical islands of capacity. Virtual segmentation is useful for private cloud implementations where a large pool of compute capacity is used, but a smaller unit of capacity is assigned to a business unit and the resource entitlements.
DRS will also act as the governor for the entitled resources. As virtual machines are added to the environment, DRS will ensure that the new VM will not be placed on a host that will violate the resource pool policy. Additionally, as utilization changes across the VMs, DRS will rebalance the cluster while ensuring the resource pool policy is not violated. The result is improved overall utilization and a simplified infrastructure because the administrators do not have multiple environments to support and maintain.
Summary
DRS does a tremendous job ensuring that VMs are always running on a host that is able to provide the resources demanded. However, it also enables the vSphere administrator to simplify operational tasks which lowers manageability or ensures compliance within the organization. Without DRS these operational tasks take more time to complete lowering the productive utilization for the operations staff. With DRS, the end result is happy admins!
