Introducing the new DRS logic and UI including VM DRS scores

The first release of Distributed Resource Scheduling (DRS) dates back in 2006. Since then, the data center and workloads have changed significantly. With the latest release of VMware Cloud on AWS, revision M9 , we introduce the new DRS UI that accompanies the improved DRS logic that was already introduced in release M5.

The new and improved DRS logic is more workload-centric rather than cluster-centric, as it was before with DRS. DRS is completely re-written to have a more fine-grained level of resource scheduling with the main focus on workloads. This blog post goes into detail on the new DRS algorithm, and explains how to interpret the metrics as seen in the new UI.

The Old DRS

vSphere DRS used to focus on the cluster state, checking if it needs rebalancing because it could happen that one ESXi host is over-consumed while another ESXi host has less resources consumed. If the DRS logic determined it could improve the cluster balance, it would recommend and execute a vMotion depending on the configured settings. That way, DRS used to achieve cluster balance by using a cluster-wide standard deviation model.

The New DRS

The new DRS logic takes a very different approach. It computes a VM DRS score on each host and moves the VM to the host that provides the highest VM DRS score. The biggest difference to the old DRS version being that it no longer balances host load. But now, we improve the workload balancing on the cluster by focusing on the metric that we care most about, the virtual machine happiness.

VM DRS Score

The new DRS logic quantifies virtual machine happiness by using the VM DRS score. First, let me emphasize that the VM DRS Score is not a health score for the virtual machine! It is about the execution efficiency of a virtual machine. The score values range from 0 to 100% and are divided into buckets; 0-20%, 20-40%, and so on.

Obtaining a VM DRS score of 80-100% indicates that there is mild to no resource contention. It does not necessarily mean that a virtual machine in the 80-100% bucket is doing way better than a virtual machine in the lower buckets. That is because there are many metrics that influence the VM DRS score. Not only performance metrics are used, but capacity metrics are also incorporated in the algorithm.

The performance drivers for the VM DRS score are contention based. Think about; CPU %ready time, good CPU cache behavior and memory swap. The reserve resource capacity, or headroom, that a current ESXi host has is also taken into account to determine the VM DRS score. Will the virtual machine be able to burst resource consumption on its current host and to what level? Are there other ESXi hosts in the cluster that have more headroom available? All these factors play an important role in the calculation of the VM DRS score.

The improved DRS is no longer thinking about the relative load between ESXi hosts in a cluster, the main focus is on the happiness of the workloads. Next to VM DRS Score, DRS presents the Cluster DRS Score in the UI. It is calculated using an aggregation of all the happiness VM scores in the cluster. DRS will try to maximize the execution efficiency of each virtual machine in the cluster while ensuring fairness in resource allocation to all virtual machines.

The vSphere cluster summary overview provides insights on what is happening from a DRS perspective.  If you require more information on VM DRS Score, the new UI will provide that information to you.

View All VMs

When clicking on the ‘View all VMs’ option in the cluster summary DRS view,  you will be presented an overview of all virtual machines in the cluster and more detailed information about their resource claim and entitlement.

There might be situations where you will see a lot of CPU stress, so a high number of CPU %ready time in the CPU Readiness column. Or a high number of swapped memory. Those are clear indicators that the workload utilization has possible depleted the cluster compute resources. You can use this information to move workloads to other clusters, or to scale out the cluster resources by adding additional ESXi hosts. The latter we can do automatically in VMware Cloud on AWS usingElastic DRS.

UI Walkthrough

Find the new UI in vCenter Server. When you click on cluster summary overview, the new DRS UI is shown on the right-hand side by default. Expand the DRS view to get immediate insights. Check out this UI walkthrough to get an understanding of what it looks like in vCenter Server.

To Conclude

We are pleased to launch the new DRS UI in the M9 release ofVMware Cloud on AWS. The improved DRS logic has proved to be a big improvement when it comes to virtual machine resource entitlement. DRS helps workloads perform as optimal as possible!

Future blog posts will provide more detailed information about the new DRS logic. Stay tuned!