Home > Servers > Rack and Tower Servers > Intel > White Papers > The science behind the report: PowerEdge R750 using Kubernetes containers > Introduction
We deployed two on-premises Tanzu Kubernetes Grid (TKG) clusters to a Dell EMC PowerEdge R750 to run the Weathervane 2.1 application- level performance benchmark and determine the number of simulated users (WvUsers) that the hardware can support without violating the built-in Weathervane quality-of-service (QoS) requirements. One cluster hosted the Weathervane benchmark application instances, and the other hosted the Weathervane workload drivers. We sized the two clusters based on the Weathervane user guide’s application resource requirements configuration sizing tables for the small2 application instance size. We also deployed a third TKG management cluster as part of the TKG setup process. We deployed VMware vCenter Server® 7.0 U2 and a run-harness auxiliary VM to a separate infrastructure host to run the benchmark and manage the testbed.
Our application instance cluster consisted of a control plane VM with two vCPUs, 4 GB of RAM, and 20 GB of local storage, as well as 18 worker-node VMs with four vCPUs, 32 GB of RAM, and 200 GB of local storage each. Our workload driver cluster consisted of a control plane VM with two vCPUs, 4 GB of RAM, and 20 GB of local storage, as well as 10 worker-node VMs with four vCPUs, 24 GB of RAM, and 50 GB of local storage each. The required TKG management cluster consisted of a control plane VM with two vCPUs, 4 GB of RAM, and 20 GB of local storage, and a single worker VM of the same size. These cluster sizes allowed us to host 11 simultaneous Weathervane small2 application instances on the server to demonstrate the maximum number of simulated WvUsers the system could support across all application instances.