The average edge infrastructure is growing at an exponential rate as more devices are being utlilised within organisations across various sectors.
“By 2021 there is estimated to be between 25 billion (Gartner) to 31 billion (IDC) connected edge devices, said Stefano Maffulli, senior director of community and marketing at Scality. “Imagine mini data centres pushing information back to stadium owners about how long people sit in their seats at a sporting event, or video images stored and managed for city officials at a traffic light.
“This is already happening. Endpoints or edge devices are constantly collecting data, in hospitals, sports stadiums, airports, ships, retail stores, offices and more.”
Considering this, it’s vital that edge infrastructures and devices are managed regularly in order to gain consistently optimal performance. But, how can this be achieved? Experts in edge technology provided their insight into this matter.
Start with the software
Firstly, according to Dale Kim, senior director of technical solutions at Hazelcast, edge infrastructure management must begin by addressing the software throughout the network.
Why edge computing is forcing us to rethink software architectures
“A cost-effective approach to optimising network bandwidth usage starts with the software,” said Kim. “Adding more, faster hardware is certainly always an option for addressing networking challenges, but not a cost-effective one.
“Understanding the data flows driven by the software can often help to attain end goals such as higher application performance without increasing hardware investment. This is especially true in edge computing environments and other distributed systems that entail delivering large volumes of data from remote sources to a central data centre.
“Even today’s modern architectures based on microservices, containers or Kubernetes, which necessarily require significant network usage for internode communication, can be enhanced to reduce bandwidth usage.
“In addition, application performance can often be dramatically improved by reducing the latency of reads and writes to non-volatile memory like hard disk drives and solid-state drives. In these situations, network usage might be acceptable as is, so the optimisation effort focuses on reducing the amount of slow storage accesses.”
Discovery processes
Another aspect to consider when managing edge infrastructure and devices is to invest in discovery processes.
“Edge by nature creates a distributed approach – accelerated by the current global pandemic – that needs a more flexible style of management,” said David Shepherd, area vice-president, pre-sales EMEA at Ivanti. “But ultimately, if we don’t know what we are managing then it becomes difficult to even start managing in a comprehensive manner.
“Effective discovery processes allow an organisation to apply the right management policies at the right time. As more devices start to appear at the edge, the context of the device plays a crucial role.
“This includes the type of device and the interaction it has with the infrastructure, plus its location (often remote). Understanding what a device is and how it interacts is again crucial to applying a comprehensive management approach.
How to ensure edge device security
“Discovery is more than just knowing what you have, it’s also about understanding what type of interactions are occurring and where they are taking place. It’s is an ongoing process, and without a continuous discovery motion then what is perceived to be known soon becomes unknown, and you must be confident that you know 100% of your infrastructure. Effective discovery can identify all the connected devices, from multiple sources and give you the visibility and confidence to deliver comprehensive management with 100% accuracy.”
Edge analytics
With the growth in edge devices occurring within companies across the world, the amount of data generated is also increasing. This calls for an analytics system to ensure accurate visibility, which can aid efficiency.
Alan Jacobson, chief data and analytics officer at Alteryx, explained: “Performing analytics at the edge is all about moving the computational controls and the processing of the analytics as close as possible to the source of the data. Historically, edge analytics has been viewed as critical to the management of edge infrastructure and devices for several reasons.
Rebuilding your data analytics capabilities in a post-Covid world
“Firstly, lack of external network connectivity to a device making it necessary to process data at the edge. Typically, this has been due to difficult environments or security requirements. Secondly, a need for speed that prevents sending data through a network due to latency, where moving the data costs more in terms of time than having the processing power of a data centre or the cloud available.
“This is absolutely true for certain use cases. On the factory floor, for example, there is a desire to prevent network connectivity from bringing an entire plant down. In fact, in many factories, the level of bandwidth currently available can often be too low to have all equipment sending data back to the data centre. In this case, it is critical to place analytics tools at the edge with no disruption, sitting the algorithm next to the hardware.
“However, for businesses with non-critical use cases, this is changing. Over time, the drivers behind the need for edge analytics have changed as network speed and connectivity become faster and more prevalent. As such, the roundtrip of data to the network – which is going faster every day – will not hinder digital progress and thus businesses are increasingly happy to manage infrastructure and devices in this way.”
Automation
Lastly, edge infrastructure and device management can also benefit from automation technology, with this aiding the aforementioned analytics tools.
“Automation is extremely relevant when we are looking at the management of infrastructure and devices at the edge,” said Janet Liao, principle product marketing manager APAC at Talend. “Automation massively enhances the process by not only capturing data at the edge, but reconciling, processing, and analysing it in real-time. This speeds up the end effects by removing humans from the loop except for a monitoring and auditing role.
CRM Process Automation: how to implement in your organisation
“Zero-touch provisioning, for example, enables easier onboarding of IoT devices onto an IoT cloud platform, e.g. AWS, as it enables automatic provisioning and configuration. This prevents developer error during the provisioning and configuration process, as well as provide a more secure interaction between the device and platform as the security framework had already been established on both ends during the pre-production stage.
“Yet, with analytics and automation, one must not forget quality and trustworthy data. If partial or imperfect data is guiding actionable insights in managing the edge, it is highly likely that the wrong decision will be made.
“For example, if a retailer is deciding on what products to highlight on a website homepage, but the insights driving this decision aren’t reflective of the business as a whole – new customers, existing customers, product availability, product demand – the final choice could harm the bottom line.”
