Let’s start with the reality that I have 27 pages of answers from Chat GPT and Google Gemini. I have to admit, I have a problem trusting either of them – Gemini because it’s always complimenting me for being “insightful” or asking the “crucial” question, and Chat GPT because it doesn’t have a personality and just gives me its “facts,” so I never know if it’s inherently biased or not. But, as I asked the questions about edge compute replacing the cloud, I have to say, the answers seem to come directly from the hyperscalers. At the end of this, I hope you’ll tell me what you think and who is hallucinating.
Let’s start at the beginning. Recently, I was talking to a friend who works for a company that makes servers. He was telling me how tough the business is competing with the hyperscalers. Based on my article about serverless solutions, I came to a belief that there was a point where owning your server was cheaper and more manageable than the cloud.
I know that some hyperscalers have deployed edge services, so I wanted to limit the scope. I posed the question, “Does edge compute belong inside a private network?” Everyone agreed that there were advantages to being inside a private network.
Gemini was emphatic, concluding in bold.
“Placing edge compute infrastructure within a private network offers significant advantages in terms of security, reliability, control, and compliance. While hybrid architectures involving secure connections to the cloud are common, the foundational layer of a private network provides a robust and protected environment for edge computing deployments.”
Chat GPT gave me this answer.
“In summary, deploying edge computing within a private network is generally recommended when security, low latency, data sovereignty, and operational control are critical. However, organizations should assess their specific needs, considering factors such as scalability, cost, and geographic distribution, to determine the most suitable deployment strategy. In many cases, a hybrid approach that leverages both private and public networks may offer the optimal balance of performance, security, and flexibility.”(Verizon)
At least I could consider the source.
I wanted to make sure I was free of the hyperscalers, so I asked the next question, “Who should manage it?” I was expecting to see some outsourcing suggestions that put the hyperscalers back in. That did not happen directly, unless you wanted to call them Managed Service Providers (MSP)s. Most of the MSPs I know are still primarily voice, but I accepted their answers.
Gemini gave the most complex answer pointing out several parameters to getting the right answer for a specific company.
- Scale and complexity of the edge deployment
- Industry and use case
- Security requirements
- Existing skill sets
- Budget and resources
- Organizational structure and culture
“There's no one-size-fits-all answer. The ideal management structure for edge compute within a private network depends on a careful assessment of the organization's specific needs, resources, and the nature of its edge deployments. Often, a collaborative approach involving the relevant internal teams, potentially augmented by external expertise, proves to be the most effective strategy.”
Okay, I will accept the parameters.
Chat GPT was far more literal, answering the question by telling who should be on the internal team.
- Chief Information Officer (CIO)
- Chief Data Officer (CDO)
- Edge Computing Engineers
- Edge Security Specialists
- Network Administrators
Sounds like a strong committee.
“Managing edge computing within a private network is a multifaceted endeavor that requires coordinated efforts across various departments and, at times, external partners. By establishing clear roles, fostering collaboration between IT and OT, engaging with experienced partners, and investing in continuous training, organizations can effectively harness the benefits of edge computing while mitigating associated challenges.”
Another reasonable answer.
Now we come to our difference of opinion. I posed the question, “Can edge compute replace the cloud?” Just to be clear, at this point, I have not mentioned IoT, so a crucial part that was on my mind was not understood by the AI. It was thinking that large language sets were necessary, while I was thinking of IoT solutions that require smaller language sets.
Perhaps I am overly sensitive and, yes, I realize it’s not “human,” but I felt I was being mocked by whatever input it was using.
“While it's tempting to think that sophisticated AI at the edge could completely replace the cloud and lead to significant cost savings, this scenario is unlikely to be a full replacement in the foreseeable future.”
“While AI at the edge offers significant advantages and can lead to substantial cost reductions in specific use cases, it is unlikely to completely replace the cloud. The cloud remains crucial for AI model training, global data analysis, and centralized management. The future lies in a synergistic relationship where AI capabilities are strategically distributed between the cloud and the edge to optimize performance, cost, and efficiency.”
Now remember what I said about ChatGPT – the answer to this query was a primer that made me feel I was being talked down to.
“Edge computing enhances cloud computing by addressing its limitations in specific scenarios. For example, in environments where real-time processing is critical, such as medical robotics or autonomous driving, edge computing provides the necessary immediacy. However, for tasks requiring extensive computational power and data storage, cloud computing remains indispensable.
Moreover, advancements in technologies like AI and 5G are fostering the integration of edge and cloud computing. Edge devices are increasingly capable of handling complex tasks, while the cloud continues to evolve, offering more robust services. This evolution underscores the importance of a hybrid approach, where edge and cloud computing coexist to meet diverse and dynamic computing requirements.
In summary, edge computing does not supplant cloud computing but rather complements it. By working together, they provide a comprehensive computing framework that addresses the full spectrum of modern technological demands. Organizations can harness the benefits of both paradigms by adopting a hybrid architecture, ensuring optimal performance, scalability, and efficiency.”
Now I backtracked a bit and said I was looking for AIoT scenarios that need a limited language set and got back conciliatory answers, but ones that still wanted to go out to the cloud even though it added to the security risks.
I believe I am right that the cloud may not be necessary for on-premises AIoT solutions, from asset tracking to video monitoring and many things in between.
I would like to hear from you if you think edge compute will take over some aspects of cloud services and should be designed to be on a closed network to avoid security risks. Vote however you choose but don’t let facts get in your way.
Edited by
Erik Linask
