peshkova - Fotolia


VMware edge initiatives improve management at the edge

VMware is moving to the edge. The vendor has introduced products such as Edge Device Manager and new capabilities such as running ESXi on ARM to ease edge management.

More and more data is being created at endpoints by edge devices and sensors as opposed to traditional data centers. As a result, more data is processed at or near those endpoints, and the software used to manage these endpoints must include features such as automatic redundancy and data protection to keep the data safe. VMware intends to help.

Endpoints include a variety of devices. As the number of those devices increases over time, the cost of managing them will increase as well. Organizations with many endpoint devices must deal with a huge number of management tasks, such as firmware updates and application deployments. For large organizations, it can be a struggle to maintain all their heterogenous devices and most likely will require a central management system.

VMware aims to provide methods for managing those devices remotely, with several VMware edge initiatives such as Project Pallas and ESXi on ARM. The virtualization vendor has also experimented with deploying an ESXi hypervisor, which can run several VMs on other small form-factor devices.

VMware Project Pallas

Project Pallas focuses on managing edge devices and extending cloud services to the edge. VMware's software-defined data center (SDDC) products focus on the traditional data center and the cloud, but not edge clusters. Project Pallas provides a unified platform on edge devices with isolation, enhanced security and multi-tenancy.

The main piece of software to come from Project Pallas is a VMware Edge Device Manager (EDM) that integrates with vCenter server as a plugin.

VMware Edge Device Manager
Edge Device Manager integration with vCenter Server.

EDM is a VMware virtual appliance that shares the same web user interface with vCenter server. It runs at the edge as an agent VM and maintains a connection with the main data center, even through high-latency WAN links.

Admins can access monitoring information -- such as CPU, memory and storage usage -- via EDM, and can also deploy new tenant VMs. EDM can also perform some basic VM management tasks, including starting, stopping and restarting VMs at the edge.

Organizations can deploy EDM in several types of edge devices. For example, they can deploy EDM in Hivecell devices at edge locations to manage those devices.

Hivecell devices
Hivecell devices are stackable edge devices.

Hivecell devices are powerful, low-energy server devices designed for containers and machine learning. But a Hivecell device can also run an ESXi host with a couple of VMs on it. A Hivecell can be used for many purposes. For example, it can be shipped to an edge location, where even a non-technician could easily plug it in and start it up. By stacking one Hivecell device on the top of another, an organization can extend a VMware cluster without having a highly qualified VMware certified engineer on site.

Each Hivecell device provides a 10 GbE connection, which runs through magnetic connectors, and has an x86 CPU and four disk drives. This means that an organization can run VMware vSAN by stacking just three Hivecells.

A Hivecell operates on a battery and can switch from wired to integrated wireless connection. If one of the devices breaks, it can be easily replaced; the device won't lose its connection to the other Hivecells within the vSAN cluster. Hivecells can also support VMware Fault Tolerance for zero-downtime.

ARM architecture and other small form-factor devices

Smaller devices than Hivecell -- such as ARM or Intel-powered devices -- can run VMware. These devices are fairly simple to operate and can run one or two VMs each at edge locations.

The goal with Project Pallas is to provide a unified platform on edge devices with isolation, enhanced security and multi-tenancy.

For several years, VMware has experimented with running ESXi on ARM architecture. Because ARM architecture has a great deal of processor efficiency, it makes sense to adapt ESXi to run on it. ARM architecture is also suitable for very small form-factor devices, such as Raspberry Pi, which has 4 Gb of RAM and costs only about $55.

The Intel CPUs in small compute sticks, which feature Core m5 CPUs and 5 Gb of RAM, have also been tested with ESXi. In 2019, VMware demonstrated a live vMotion over Wi-Fi using such devices. Intel CPUs can run small workloads, such as monitoring robots.

Admins can fully manage small devices through vCenter server. Intel CPUs and Raspberry Pi devices are compatible with Distributed Switch and vSphere High-Availability, which can mitigate any hardware failure by restarting a VM on remaining ESXi hosts within a vSphere cluster.

Modern edge layers

In the future, edge data should present itself in three layers. Each of these layers will operate different servers, IoT devices and small form-factor devices.

A factory SDDC is a main on-premises data center with connections to cloud data centers.

Production line clusters are local IoT devices, running some local analytics VMs and monitoring VMs. This tier transfers the most important data back to the factory SDDC.

Robot monitoring and control devices are the smallest devices. These represent sensors for monitoring robots, data capture or running low-latency applications. They transfer only key events upstream.

VMware at the edge is an emergent technology that might not find broad-scale implementation for several years. However, VMware's close work with startups, such as Hivecell, indicates that this technology will find its way to market.

However, VMware must first create edge standards that use a common firmware for different IoT devices -- firmware that comes preinstalled with VMware ESXi as well as other Linux distributions, such as Debian, Fedora, FreeBSD, Photon OS, SUSE Linux Enterprise Server and Ubuntu. ESXi on ARM isn't generally available yet, but VMware has a dedicated ARM team working toward a final product.

Dig Deeper on VMware networking

Virtual Desktop
Data Center
Cloud Computing