Close

The disruptive nature of the Edge

By ,

How has the Edge evolved?  

Since my last blog on this topic just over a year ago the move to Edge compute services has gained pace. The hyperscalers have launched a range of Edge solutions, for example Outposts from Amazon Web Services, Azure Stack Edge from Microsoft and Anthos from Google.  Analyst firm Gartner, forecast that “By 2022, more than 50% of enterprise-generated data will be created and processed outside the data center or cloud, up from less than 10% in 2019”. We’ve also seen that the vast amounts of data created by the IoT and specifically Industry 4.0 is driving the move to process data away from the cloud. However, a number of things have changed in terms of what the Edge now looks like.    

Less focus on VNFs and service chaining in networking

A year ago there was a big focus on the Edge being the place where organisations could run virtual network functions (VNF) like routers, SD-WAN, acceleration and network optimisation.  These could be service chained, so that clients could pick for example their preferred router, SD-WAN and firewall and run them in sequence. There were lots of advantages in doing this in terms of not having to ship and maintain multiple boxes around the globe. The concept of a single box has remained, but what we have seen is vendors packaging these products together themselves so that service chaining has become less important. We’ve also seen more emphasis on compute than networking, with one exception called 5G, which I talk about later.  

Making IT “green”

Customers I talk to are planning to process up to 50% of their data at the Edge, and it’s not latency that’s the primary driving force.  It’s driven by the need to reduce connectivity costs and the requirement to keep critical processes running e.g. manufacturing optimisation algorithms, even if there are network issues. Data sovereignty and things like GDPR remain important, however, one of the new concerns is around the environmental impact. Processing data at the Edge usually involves less duplication, comes with lower network energy costs, and gives the user greater control in terms of the use of clean energy, not always possible when using cloud data centres.[1]   

The threat of cyber attacks

One of the main concerns at the Edge and also one of the biggest barriers to the deployment of the IoT and Industry 4.0 is cyber security. Increasingly the Edge becomes a point of convergence between two worlds Operational Technology (OT) the systems that run equipment in factories, refineries, and mines for example, and Information Technology (IT). Industry 4.0 solutions like predictive maintenance need data from both worlds e.g. SCADA from OT and ERP from IT. The OT world is typically very vulnerable to cyber-attacks [2], 80% of which come from IT and has traditionally relied on security through obscurity and air gaps. Once you join the dots, critical processes become vulnerable because they run on old proprietary software, with poor password protection, limited patching, and no authentication.  So identifying and mitigating vulnerabilities becomes a major focus area.

Future research

So what are some of the innovative things we are doing? 

One of our focus areas is 5G, where we have a very large research programme. To get the super low latencies, high availability and vastly improved bandwidths promised by the 5G vendors you need to move some processing from the core to the Edge. We’ve been doing a lot of work on MECs, (Multi-Access Edge Computing) where 5G applications can be run at the Edge of the Radio Access Network, and even looking at how these MECs could be virtualised on an Edge compute box. Typically applications might include augmented reality training without the need for a back pack for the batteries/compute, or ultra-high definition cameras to monitor robots.

Another area is ZTO (Zero Touch Orchestration). The main technical challenge with the remote management of Edge devices used for the IOT, is the need to establish initial trust between the Edge device and management software which forms part of the IoT platform. Third party on-boarding services e.g. (Intel® SDO) can be used for attestation (i.e. to establish that initial trust), however, one of the difficulties has been how to integrate one of these technologies into an IoT platform architecture. The solution currently in development, means that when an IoT device is installed and connects for the first time; it is registered and connects automatically and securely as a bona fide and fully trusted device into the cloud-based IoT platform. The platform then adopts the device as a secure device which it begins to monitor and maintain.

So if you are transforming your Edge to deliver the applications of the future and evolve the network functions you deliver today, why not talk to us?

Andy Rowland

Categories

Let's stay in touch