Solving Edge Computing Challenges to Meet Enterprise IoT Needs

26 September 2019

The Internet of Things (IoT) is here and smart devices are now everywhere. Within the next 12 months, there will be at least 30 billion such devices linking people, businesses, and even several industries.

These devices are already making a massive impact on businesses. According to Cognizant, IoT technologies have reduced supply chain costs by at least 20% while increasing productivity by 10-20%. The technologies have also reduced time-to-market by as much as 50% in some industries. 

helloquence-5fNmWej4tAA-unsplash

Role of Edge Computing in Enterprise IoT

Unfortunately, many organizations still experience a range of challenges when implementing IoT initiatives. In a survey by Cisco, only 26% of companies described their IoT adoption initiatives as a trouble-free. The rest said the processes were “more complicated than they first thought” or “took longer than expected.” 

In a separate study by the IoT institute, almost 50% of respondents said they experienced significant challenges when implementing the technology. The problems ranged from latency to scalability. 

Edge computing has since been adopted to solve some of these challenges. As the name suggests, edge computing brings computing resources close to the information-generation source. Storage and processing of data, in particular, happens right where the data is generated. 

The result is that previous bandwidth and latency challenges become non-issues. Edge computing makes IoT not only faster but also more agile and resilient. The technology also brings additional benefits such as improved network connectivity, network security, data privacy, and autonomy.

Fresh Challenges

The problem, however, is that edge computing itself has come with new operational and design challenges. Edge computing & IoT is highly distributed. Any organization can have thousands of sensors and associated gateways. All these devices have firmware, OSs, containers, and essential software. A large number of devices, users, and edges/nodes presents several new challenges, including;

  • Differing policies and practices

The policies and practices guiding edge computing and IoT differ significantly. As such, the policies used in traditional data centers cannot be blindly applied to edge nodes for several reasons. For example, edge deployments are much more distributed and dynamic than conventional cloud storage centers. Integrating edge to IoT, therefore, isn’t as straightforward as enterprises would wish. 

  • Need for continual maintenance 

Edge computing requires regular data center maintenance operations, including; updating, monitoring, and change management. Additionally, it requires other high-level operations such as updating machine learning models and device management. If this were to be done on just one or a handful of nodes, then there wouldn’t be a big problem. Unfortunately, the maintenance has to be replicated to possibly thousands of edge nodes and clusters. It’s a truckload of work!

  • Increased security risks

Bringing the cloud and data centers to the edge of the computing device also opens new doors for cyber attacks. Attackers can target weak endpoints such as nodes and devices as entry points to valuable assets within the business network. Once the attackers gain unauthorized access to a device, they can execute more severe attacks such as a distributed denial of service (DDS) attack. It’s unfortunate that securing each node on each device can be a monumental task. 

  • Cost Growth

Finally, edge computing requires additional monetary investment, with the cost being dependent on the size of your organization. In a large organization with a vast number of devices, more nodes are required. Therefore, planning for and deploying the nodes can be costly. As the company expands, you’re also needed to add further endpoints to the network. Buying the necessary hardware and software and having professionals add new nodes is a complicated and expensive undertaking. 

Possible Solutions

Experts say that planning is the best solution to these problems. And there are four crucial factors to consider;

  • Do you need edge computing at all?

It may well be that you don’t need edge computing in the first place. Perhaps a traditional cloud solution is the best option for your enterprise. Ask yourself whether adopting edge computing would benefit you significantly or you would face more computing challenges. If not, then it would be best for the enterprise IoT to keep using the existing cloud model.

Deloitte has come up with a guideline to help organizations assess their need for edge computing & IoT. The guide is based on risk vs. reward. The following are some of the highlighted considerations;

  • Autonomy and resilience: If your project demands autonomy and resilience; if connection interruptions cannot be tolerated, then you should consider edge computing.
  • Urgency: Are you working on a project where the speed of execution heavily influences results? Are you likely to run into mistakes if you can’t complete tasks on time? If so, then you should consider edge computing.
  • Privacy and security: Sometimes, to comply with regulatory privacy and security requirements, you need data at the edge computing. 
  • Data volumes and bandwidth: Finally, you should also consider edge computing if you need to make more economical use of bandwidth. This would be especially important when data volumes are large, or some of the data you gather might be unnecessary. 

There are many other factors, including operational technology and protocols, to help you determine whether investing in edge solutions would be beneficial to your organization.  

  • Do you need that extra node? 

Once you’ve determined that edge is the solution, the next step is to consider where and when to add new nodes. Remember that every additional node creates new complexities. Therefore, you want to have as few of them as possible. 

  • What capabilities are needed at the edge?

There are wide variations in edge capabilities, responsiveness, and placement. In some situations, packaged solutions offer invaluable simplicity. The only downside is that pre-configured solutions tend to be less flexible. Self-constructing solutions are a much better option as they are more flexible. But they are more expensive.

  • Choose the right vendor

Most vendors provide hardware such as servers and gateways as well as local storage, data processing, and analytics. However, some essential functions, such as protocol handling and device management, are outsourced. Ideally, you should find a vendor that either provides everything you need or makes all the arrangements on your behalf.   

You Have a Big Decision to Make

Edge computing has a significant role to play in the successful implementation of IoT. But, edge solutions also come with challenges. You’ll need to decide whether the technology and additional nodes are worth the risk. NIX can help you analyze your organizational needs to come up with the best decision with regards to the adoption and implementation of edge computing. Contact us today for a free consultation.