To ensure an accurate comparison of the update approaches, we were rigidly systematic during our testing. In scope for the tests were hardware updates only, not updates to host and VM operating systems or applications. We applied component firmware and operating system drivers from the January 2020 release of the Support Matrix for Microsoft HCI Solutions from Dell Technologies to the newly deployed cluster. We then followed the procedures in the Microsoft HCI Solutions from Dell Technologies: Operations Guide for Managing and Monitoring the Solution Infrastructure Life Cycle to bring the cluster into compliance with the July 2020 release of the support matrix. We used the following test scenarios for comparison:

• Applied hardware updates following a fully automated, cluster-aware updating approach enabled by the Dell EMC OpenManage Integration
• Applied hardware updates following a manual, node-based approach enabled by the integrated Dell Remote Access Controller (iDRAC) on each AX node

The following table lists the components that required an update, the type of update, and the starting and ending release versions. Creating a 6-month gap between the release versions offered a realistic scenario, considering the complexity and time commitment that is typically required to update the hardware of most traditional virtualization environments. This gap also ensured that multiple firmware and driver packages would require updates to bring the cluster into compliance.

Table 3.              Updates required during testing

Note: The Broadcom NetXtreme Driver Family drivers were not installed during initial deployment of the cluster. The Broadcom adapters were not in use with the fully converged network configuration.

We performed the update process first by following the automated approach. Then, the firmware and drivers were reverted to the January 2020 release level before we used the manual update approach. No errors or failures requiring investigation or remediation occurred in either the automated or the manual scenario. 

The following criteria were common to both testing approaches:

• The July 2020 release of the Support Matrix contains the Dell Update Packages (DUPs) from the Dell EMC June 2020 block of updates.
• Although five VMs were provisioned to each cluster node using VMFleet, no load was generated using VMFleet during the update process. The goal was to have VMs running on each node, which would require a live migration when the host was placed into maintenance mode.

The following information applies to the test procedures and results discussion in the remainder of this paper:

• Firmware and driver installation packages are referred to as Dell Update Packages, or DUPs, throughout the update process steps.
• For task durations, we distinguish between attended and unattended administrative time as follows:

• Attended—Requires IT operators’ focused attention to complete.
• Unattended—Requires only periodic progress monitoring by IT operators. Operators could potentially do other work during this time.

• Mistakes that occur during the operation of an HCI cluster could have catastrophic consequences to the business. In the following sections, the update process tables—Table 4 and Table 5—include a “Human error quotient” that identifies potential impact based on a relative scale of None, Low, or High. The subsequent discussion in Human error quotient provides additional information.

Many variables can affect update performance and duration. This paper includes specific cluster configuration and update process details to ensure an accurate comparison. Other variables that could affect the update performance are:

• Number of VMs running on the cluster
• Types of workloads running in the VMs 
• Number, type, and version of the updates to be applied
• Hardware configuration of the nodes
• Network configuration
• Available Internet bandwidth for downloading all applicable component DUPs