Nowadays, many organizations that adopted the cloud are looking into the Edge as a natural extension for their cloud-based solutions. On the other hand, the ones at the very beginning of the Cloud journey are way more aware of the Edge and the Cloud, so they are considering the usage of both technologies at the very beginning.
So, the questions are:
Is the Edge something that will replace the Cloud?
Or Edge usage over the Cloud will bring back the old times; on-premise development?
Since there is not a single good answer for that, I will give the pros and cons of such an approach in this article and try to answer a question from the title.
What is the Edge exactly?
The Edge has multiple meanings, of which two are most accurate. Namely – the physical device controlled from the cloud level, but it is being placed on site, close to the data source or process. The second meaning is a software framework that allows the implementation of needed functionality and allows the management of this software ecosystem securely.
Is the Edge the same as the Cloud?
Edge Computing is an application that uses ie. computer vision and data processing techniques done close to the data sources. Cloud computing aims for similar applications but done at much bigger scale (and much, much more). Edge can interact with devices in real time because there is no Internet along the way – unlike the cloud.
The technical perspective of Edge Computing
As said, Edge is a physical device with dedicated software installed on top of it. Depending on the Edge operating environment, it should meet different expectations. For example, in the industry environment, it should be built on top of the industry-class PC (it should have increased resistance to humidity, dust and vibrations), while when using it in an open healthcare or well-being environment, a smart band or a standard mobile phone is enough.
It is worth mentioning that to provide a good level of security in IoT solutions, Edge should always use Cloud Native protocols secured with CA-signed certificates and it should support hardware security (HSM – hardware secure module, an example of implementation of this standard could be a TPM – trusted platform module).
What is a business perspective for Edge?
From a business perspective, Edge is an enabler, something that can address challenges commonly existing in the IoT world. Examples of such use cases can be found below.
A. Protocol translation
Have you ever wondered how to connect to our platform on a device that does not speak “cloudish”?
Or is it so old that it has no idea about internet connectivity?
Well, someone did, and the answer is Edge! It, in this use case, can work in multiple ways, but to keep it simple: Edge is a kind of gateway that, from one end, connects such a device and reads the data from it, and on the second end, it sends the same data (or maybe just little bit of processed data) to the Cloud using Cloud Native protocols like AMQP or MQTT. Of course, such a translation can go in different flavors – data can be associated with a device entity or it can abstract all device entities and act as a data producer – the choice is yours (well, if you are an architect).
B. Edge Computing
Sometimes, it is necessary to preprocess the obtained data before sending it to the Cloud. For example, the data from accelerometers. There are thousands of measurements per second in processing such raw data. So it makes no sense at all to send it, in such a raw form, to the Cloud. In this scenario, some calculations and pre-aggregations are done (avg, std. deviation, etc.)
C. Offline scenario
This is a handy use case, especially in domains like production, marine or autonomous vehicles. Edge by idea is supposed to communicate with the Cloud, being updated from the Cloud level, etc. In a perfect world, everything should work as designed but… There are still organizations that are far away from reliable internet connectivity. In the marine domain, it is quite easy to imagine that there are still places without network connectivity in the ocean.
If the Edge is used only for data acquisition or some data preprocessing, it is not a problem. Edge is buffering the data (where the only limitation is a hard drive of the Edge device), but:
What if some algorithms calculate, i.e. sailing speed, taking into consideration current ship state and context information like a weather forecast?
In such scenarios, it is wise to keep these workloads as close to the supervised by the Edge devices as possible. Suppose network connectivity to the Cloud is broken. In that case, the Edge can still be operative, ensuring production lines run smoothly and performing calculations for potential issues (such as overheating of a welding robot) even without access to the cloud platform. This use case is strongly related to Edge Computing itself.
Does the Edge extend the Cloud, or does the Cloud extend the Edge?
If we think about this question super pragmatically, the answer is simple – the Cloud was before the Edge, so the Edge must extend the Cloud. Live scenarios show that being 100% pragmatic here is not always applicable. The honest answer is (my favorite one) – it depends!
It depends on what you may ask, the architecture or use cases of the Cloud and the Edge usage. You can say that if all major workloads, predictive maintenance, long-term storage, etc. are being run on the Cloud and the Edge is only for data aggregation, simple preprocessing and protocol translation, you can honestly say that the Edge extends the Cloud.
On the other hand, if all major workloads, storage, etc. are being run on the Edge and a Cloud is only a synchronization tool that allows you to compare all your factories’ KPIs, OEE, production efficiency, provides a data science environment and some aggregated visualization, you can honestly say that Cloud extends the Edge!
As you can see, the answer is not so simple and requires a deep Edge knowledge about the solutions it affects, so we need to be careful here.
Will the Edge replace the Cloud?
Whether Edge computing will replace Cloud Computing is a topic of ongoing debate and evolution within the world of technology. Both Edge and Cloud Computing serve distinct purposes and have their strengths and weaknesses.
Firstly, we should recognize that Edge Computing technology is not interchangeable and cannot replace others. Edge computers process time-sensitive data, while Cloud computing focuses on more asynchronous processes based on data.
It’s essential to add that Edge and Cloud Computing are not necessarily in direct competition; instead, they can complement each other in a hybrid architecture.
These two paradigms will coexist and be integrated to meet the diverse needs of the digital landscape. The future will likely witness these technologies’ continued evolution and refinement to create a seamless, efficient, and responsive computing ecosystem.
Does it make sense to run the Edge without Cloud Computing power?
Here, the answer is a bit easier than the previous question, but still not as straightforward as you might think.
Of course, the Edge can cover all scenarios in the various domains, and in such scenarios, it can be considered a standard on-premise server hosting a custom solution. Similar to the Edge with Cloud connectivity, we can use marketplaces to run some ready components and run them as containers in our solution microservices ecosystem.
So far, so good, but the real question is whether to use a cloud-based solution vs. an on-premise solution.
There are tons of articles about that topic which you may have already read or watched, so I’ll give you a few examples – high availability, storage redundancy, easy resource provisioning, or super easy deployment on multiple Edge devices based on predefined metadata (tags like device type and model, location, company, soft version etc.).
The answer to the question:
Does it make sense to run the Edge without the Cloud? – is yes.
But how it looks for real?
Running Edge without the Cloud makes no sense from a pragmatic point of view.
There is a resource consumption overhead for a framework we won’t fully use. Instead, I recommend buying Edge-type hardware and just putting our custom solution there, which will match all our performance expectations.
Edge Computing vs. Cloud Computing benefits
Cloud is designed for scalability, and running all the resource-heavy workloads on the cloud makes perfect sense since we have (almost) “limitless” resources to use. Edge computing is designed for something else – its purpose is to make calculations close to the data source, being connected physically to it in major cases. Thanks to this Edge can react way faster to potential issues than a cloud, which needs to use the internet as a connection medium. On the other hand, in a simple scenario where we want to train an ML model for overheating detection and use it against the data from our devices – it makes perfect sense to build and train a model on the cloud side while deploying it on the Edge.
The choice between Edge and cloud computing is rather virtual since both address different use cases, i.e.: big data, complicated computing, rendering is the job for the cloud while covering offline scenarios, doing the statistical analysis of the data stream, or translating legacy protocols into Cloud Native protocols are the tasks for the Edge. Many organizations are adopting a hybrid approach to leverage the strengths of both paradigms
So, is the Edge a new Cloud or not?
I will answer this from multiple angles. Each aspect is crucial and determines the answer because, for example, let’s define the Cloud. This is an abstraction for resources and services. From this perspective, can we say that the Edge is a new Cloud – well, yes! If you put an abstraction on everything that’s going on and what is computed on your device, it can be considered a private cloud with limited access.
Let us go further – what are the cloud-based service models?
IaaS, PaaS and SaaS – what is moving from the IaaS to the PaaS?
According to Marc Russinovich, this transition is called innovation (I’m thinking in the same way)! So, from this perspective, can we say that the Edge is a new Cloud? No!
Defining innovation in the same as in the Cloud, the Edge is nothing else like another on-premise server with some custom software running on it.
Let’s continue – what about the Edge as a Cloud-Managed framework? The answer is “hell no!” You cannot manage something newer, by definition better and more awesome with something old, legacy or even obsolete in this context – this would be shooting yourself in the foot!
And what about hardware?
Can it be considered (Edge enable device) as a new Cloud? Let me answer like this – I’ve already seen that somewhere. Isn’t it back to the on-prem servers and custom development in this context?
So, as you can see, there is no simple answer to this question because, just like in AI, everything depends on the context in which we ask it. But I don’t want to leave you hanging. There is an answer that might be interesting, and the theory behind that question actually aligns with real-life scenarios.
The Edge should be considered as a natural extension of the Cloud, just as the Cloud is a natural extension of the Edge!
In IoT domains, there are “somewheres” where you cannot skip Edge usage, and there are also “somewheres” where the Edge is an unnecessary cost – everything depends on the use case and should match the client’s needs and expectations. This is the true reason why cloud architects earn so much – the need to decide where and when use which service from a few thousands of Cloud services portfolio.
I hope you will find something interesting in this article that will make your life easier, or at least you will be able to impress someone during a small coffee talk.