We are in the midst of a digital transformation era. More than ever before, enterprises of all sizes are investing in data-driven artificial intelligence/machine learning (AI/ML) processes that help create efficient business operations, discover new revenue opportunities, and gain competitive advantages. Initially, organizations turned to public cloud to make sense of their data. Public cloud provides high availability and rapid scaling of compute resources for big data AI/ML applications while reducing infrastructure complexity. However, with the explosion of data at the edge, a new problem emerged — getting the data to the cloud.
For many organizations, most of their data comes from the edge — remote or branch offices (ROBO), manufacturing facilities, retail stores, restaurants, oil rigs, vehicles, weapon systems, humanitarian outposts. The data itself is produced by the ever-increasing number of sensors and Internet of Things (IoT) devices deployed at edge locations. The challenge is that for many use cases, we need to have immediate actionable insights from this data. Sending vast amounts of sensor data for processing to a remote data center or public cloud is not always efficient or feasible. And this is where the need for edge computing comes in.
With edge computing, we are going back to the future of placing compute close to where data is created, like we did in the era before the cloud. Per 451 Research report ”The Edge to Cloud Continuum“, edge computing will become an increasingly important part of the enterprise digital strategy and become a critical capability for organizations that:
- Must extract value from vast volumes of data quickly
- Need to reduce amount of data transferred over a network
- Looking to improve data security
- Must retain data locality for regulatory reasons
Some of the key verticals and use cases for edge computing include:
- Telecom – By bringing compute into the network and enabling applications to be hosted closer to the end-user, edge computing delivers ultra-low latency processing to effectively utilize 5G networks.
- Retail – Machine vision systems combine camera sensor data and AI to produce real-time actionable insights such as identifying mis-scans and ticket switching, mitigation of organized theft and shoplifting, and detection of “sweetheart” dealings.
- Transportation – Real-time AI/ML analysis of railcar vibration, laser and LiDAR sensor readings allow engineers to identify broken wheels, brake issues, dragging chains, open doors and other concerns which could lead to a derailment.
- Govt – Federal teams, from FEMA managing disaster relief, to the Department of Justice working on law enforcement programs, to the Department of Defense managing military operations, collect and process increasing amounts of sensor and IoT data at the edge, in harsh environments.
- Manufacturing – Quality control, overall equipment effectiveness, predictive maintenance, yield optimization, enhanced logistics and digital twins are among the most common manufacturing edge use cases.
- Offshore Oil and Gas – AI systems at the edge allow for safer and more efficient operations, insights into seismic activity, reservoir evaluation, faster drilling, consistent well delivery and predictive maintenance for drilling facilities.
Outside of AI/ML data science use cases above, there is a variety of mission-critical workloads that an organization may also want to run at the edge. These include automation control systems, hyper-converged infrastructure (HCI) software such as VMware vSAN, virtual desktop infrastructure (VDI), point of sale (PoS), digital signage, video surveillance, data aggregation/compression and many more. There may be a need to run a legacy operating system on a bare-metal server and at the same time have a virtualized environment with a variety of virtual machines (VMs), and then also have a compute node with graphic processing unit (GPU) support for data analytics. The challenge is running all these workloads at the edge, where there are space constraints and, often, exposure to extreme temperatures, dust, humidity, and vibration. There are several ruggedized server systems on the market. However, none of them offer the compute density and form factor required to accommodate the demands of today’s edge.
What we see in the market right now is a need for dense and modular small form factor systems that can easily be configured to run a variety of workloads. Dell has answered the call by expanding their popular XR line of ruggedized servers with a unique product that will address all these challenges. The new product will be unveiled later this year. Dell is promising to deliver a high-performance, multi-node edge server, purpose-built for ultra-short depth, low power, with modular sleds. Dell has hinted that the new XR line will allow customers to create a self-contained 2-node vSAN cluster instance to accommodate VMware vSAN requirements in a footprint not much larger than a shoe box. This all sounds very exciting, and we can’t wait until the official announcement!