A cluster is a collection of ESXi hosts and their virtual machines, where resources are shared and the management interface is common; a DRS cluster is simply a cluster with the vSphere Distributed Resource Scheduler (DRS) enabled service. A DRS cluster allows you to distribute and balance the physical resources (CPU and memory available to it) between the various VMs.
When you insert an ESX host into a DRS cluster, the host resources become cluster resources. All within a DRS cluster it is possible to manage resources in the terms described below.
Initial positioning - the moment a VM is found within the cluster, the DRS service places the VM optimally within a host, applying the recommended settings.
Balancing load - the DRS service continuously monitors the distribution of memory and CPU resources between hosts and virtual machines. Load data is continuously compared with an ideal use of resources defined by the Migration Threshold value.
Power management - by enabling VMware Distributed Power Management (VMware DPM), an optional feature of VMware DRS, it is possible to have savings at the level of energy consumption by consolidating the loads during periods of lower resource commitment. In practice, the VMs are migrated to a number of hosts that reserve the resources, the other hosts that provide surplus resources are sent on standby.
Totally there are three major elements here:
- Network (If you are using vSphere 6.5 or later versions)
In DRS point of view, your workload is the combination of CPU, Memory, and Network, but there are some considerations in term of triggering a DRS such as VM placement or host load standard deviation.
DRS does the initial placement in these steps:
1- DRS collects the list of hosts which are suitable based on available resources and prioritize them.
2- It decides which host has the highest score to act as a new home for the intended VMs.