DRS automatically balances workloads across a VMware vSphere infrastructure, with myriad benefits to server performance and power usage. The tool advances hardware resource utilization to a degree that was unthinkable back in the Pleistocene (also known as the 1990s). DRS can finesse load balancing further with some input from the VMware admin.
VMware's Distributed Resource Scheduler (DRS) is an integral part of many vSphere environments, so much so that admins may take it for granted, with simple configuration and operation. But there is a catch: Not every vSphere edition includes DRS.
What VMware vSphere DRS does
VMware describes vSphere DRS's role as aligning "resources usage with business priority by automatically load balancing across hosts and [optimizing] power consumption by turning off hosts during lower load periods." DRS aids in initial virtual machine (VM) placement, load balancing and keeping our data center as green as possible.
VMware DRS performs initial VM placement on a host. A new VM will exist on a host that belongs to a cluster of ESX/ESXi servers. DRS examines the load on different hosts running in your cluster, then selects an appropriate host on which to start up a VM. For example, if you have three hosts -- A, B and C -- with hosts A and B using most of their resources, vSphere DRS will start the next VM on the underutilized host C.
DRS continuously monitors the cluster's resources available for hosts and makes running-VM placement recommendations to balance and best utilize these available resources. For example, on the cluster with hosts A, B and C, all the VMs use similar levels of memory and CPU resources. If hosts A and B are each running five VMs and host C is only supporting two VMs, DRS will likely recommend that one VM migrate from host A and from host B onto C. This will leave hosts A, B and C each supporting four VMs, balancing the load. Depending on how you configure it, DRS can simply suggest migrations or actually initiate them. That is entirely up to you.
Distributed Power Management (DPM) is a configurable option within DRS. When DPM is enabled, DRS evaluates current and historical loads on a cluster's hosts and either automatically places hosts in standby mode until demand grows or recommends that the admin do so. With DPM enabled in DRS, vSphere can actually reduce both power consumption and cooling requirements in your data center. Think of DPM as a green technology.
DRS requirements: Not all vSphere editions
As of April 2013, VMware won't include vSphere DRS in all licensing packages. To obtain DRS functionality, you must license vSphere Enterprise or vSphere Enterprise Plus. VMware omits DRS from vSphere Standard, Essentials or Essentials Plus kits. By limiting DRS availability to the more advanced (and pricier) vSphere packages, VMware may be preventing smaller IT shops from adopting DRS. If you're on the fence, evaluate DRS in trial mode to determine if the benefits are worth the vSphere edition's price for your particular environment.
To enable and use DRS, you must have two or more hosts in a cluster, and they must be vMotion-enabled. DRS can't load balance VMs or make initial placement recommendations for starting up VMs with only one host to choose.
The VMs' disks that DRS operations will affect must be on storage that is accessible to the hosts within the cluster. Otherwise, you will not be able to migrate VMs between hosts.
How to configure DRS and run the feature
Configuring DRS is very simple. Once you have a cluster configured in vSphere, right-click on the cluster object then select Edit Settings. A menu box will offer the option to Turn On vSphere DRS. The vSphere DRS menu item offers additional DRS options, such as Automation Level (Manual, Partially automated, Fully automated and Migration threshold).
More considerations with DRS
Could DRS multiply your licensing reqs?
The Hyper-V equivalent to DRS
VMware provides brief and clear explanations of each option. But you need a good understanding of your unique environment to know which setting will work best.
In addition, vSphere has more granular DRS settings that can be manually configured, such as affinity rules, found under vSphere DRS, Rules. Affinity rules let you keep certain VMs running on the same host, for example when you know particular VMs communicate with each other a great deal. Anti-affinity rules will force VMs to run on different hosts. Use this for availability; for example, if one host fails, the anti-affinity VM on the other host will not be impacted.
To easily view your current DRS configuration in the vSphere Client, select your cluster, then select the DRS tab. This displays your Cluster properties, as well as Recommendations, Faults and History. To generate DRS recommendations, select Run DRS, which will produce a list of recommendations that you can implement by selecting Apply Recommendations. History shows your past actions. I recommend initiating a Run DRS any time you have made significant changes to the vSphere environment, such as adding, removing or migrating numerous VMs. This is a proactive way to keep clusters running efficiently.
DRS is a well-designed, effective vSphere tool that can help you optimize the use of your hardware resources. If you meet the vSphere edition and hardware requirements, try out DRS.