About Cloud, AI, IOT and other trends in our industry – a developer’s perspective.

The world of computer games and movies is becoming a reality. When I last visited Las Vegas for a conference at which I was a guest speaker, it suddenly hit me. I was walking around in the movies and games that I watched and played when I was a kid. Billboards with high-resolution LED displays that show what there is to do and where you should go. A computer in your hand, leading the way; your airline check-in on your hand-held computer, and even billboards that interact with you based on perceived emotions. We are living in the world envisioned 20 years ago and we entered this world without noticing it.

It’s amazing to see how artificial intelligence has become more and more integrated into our lives. The billboard I just described, watching your expressions and recognizing whether you like the ad or not, and quickly responding to you with other content that might please you, based on the recognized race, gender, age and skin-color. It’s simply amazing, but if you think about it, it’s also a bit scary sometimes. Las Vegas is the perfect example of how technology can be used to persuade you to buy new shiny things and watch a show, we don’t have to look far to see how cloud computing, IOT and machine learning are infusing our lives with new ways of interacting and communicating. Take the simple example of the parking warden doing his rounds. Instead of walking around with his notebook, he is now driving around with a big camera mounted on his car, taking pictures of every parked car while moving and automatically fining those who have not made their payments. That is machine learning at work. The rise of the robots, but they don’t look like robots. It is just code at work.

The rise of the cloud DevOps

When you’re a software developer, cloud computing is something you just cannot avoid thinking about. If you don’t know the cloud, if you’re not learning the cloud, chances are you will not be relevant anymore in just a few years. For businesses it’s no longer a question of whether they need to move to the cloud. It’s more about which workloads will run in the cloud first and what will be the next step. In my work as a consultant, I get to talk to many customers, and when you look at all the conversations, the cloud is the predominant conversation, besides other ways of improving the speed at which we can deliver new solutions to market.

This is where organizational concepts like DevOps meets the Cloud. DevOps and cloud computing accelerate each other and help you move faster than ever before. This has major implications for your skills as a developer. In the not too distant future, there is no hand-off between different silos in the organization that have been optimized to utilize their resources. You will see that yesterday’s organizational boundaries will cease to exist. You as a developer or IT operations professional will be responsible for managing everything yourself in production. This means no long process of getting approval. Instead, it means empowered teams can do all the work themselves. No long waits to get a server but infrastructure, configuration and security as code. That is the way forward.

The cloud and containers

When you move to cloud computing, customers often want to select a cloud vendor first before they get started. Of course this is based on their old habits of needing to choose once wisely and then sticking to that choice. That is no longer necessary. Since we don’t have to make huge capital investments in our datacenter anymore, who cares if you first start with IaaS and later decide to go for a PaaS solution? Get started and go fast, that’s more important. Just adjust along the way. Yes, you might lose some investments here and there, but your loss will be much bigger if you don’t make the jump fast enough and your competitor did. Container technology is another thing that can help you. This technology is used by all loud providers and enables you to move between them without the need for a rewrite. This gives you all the flexibility to move between cloud providers plus it gives you much better utilization of the cloud infrastructure you use. Containers are here to stay and this is the move everyone will make. You can compare it to the revolution virtual machines brought to the IT industry. With containers we even go beyond this, and we now have a universal way of building and packaging our applications and running them anywhere where they support containers. And that support is everywhere! Today it’s still very hip and cool if you create and manage your own container clusters with orchestrators like Kubernetes. But the cloud providers are going to abstract this away for us in the blink of an eye. All major cloud providers, i.e. Amazon, Microsoft and IBM, are all betting heavily on containers and the Kubernetes orchestration engine. They are all creating fully managed clusters for us that have PaaS characteristics, making it easy for us developers, because we don’t have to think about the underlying hardware platform. This is a major shift, further into configuration and infrastructure as code. This trend is rapidly expanding and something we will look back on in a few years as the war of the cluster orchestrators. We will think of it as a “tabs versus spaces” discussion. Yes, it is fun to debate, but it really does not make a huge difference, as long as we pick something. The industry has picked the winner and clearly that is Kubernetes.

Cloud, IOT and the Edge

You can’t ignore the fact that we’re getting more and more connected devices. Those connected devices make up the so- called edge of the cloud. And it is the billions of mobile devices and millions of PCs that make this edge extremely powerful if we would leverage all that computing power to do smarter things rather than just calling into the cloud network. This is where you see more and more smarts coming into our devices with new computing power. A new collective processing power provided by Graphical Processing Units (GPUs), Neural Processors Units (NPUs) and Field Programmable Gate Arrays (FPGAs) in addition to the classic CPU. All this new silicon provides dedicated computing power for various jobs that can be done at the edge to make the impossible possible.

Imagine you are running a vision recognition program to look at the goods you process and do a final visual inspection. With a machine-learning model in the cloud you need to take a photo, upload it to the cloud, and then receive an answer. With a CPU and a local machine-learning model (e.g. an ONNX Model, supported on all Windows versions these days), you can process these images much faster, even hundreds per minute. Now put this model in a Neural Processing Unit or an FPGA and you can do this a hundred times faster. All of a sudden you realize that the futuristic vision of real-time video stream analysis has already become a reality. Creating a model requires a lot of data and is something that takes time and effort. This is typically a workload you would do in the cloud itself. But exporting this model to something that can run on your Raspberry PI creates a whole new world of opportunities. The cloud, IOT devices, and Edge computing are the underlying pieces of the puzzle that are turning machine learning into something that’s more and more real-time and sophisticated. Only then we might have achieved real artificial intelligence, as opposed to the glorified machine learning that is called AI nowadays.

New client frameworks to interact with our systems

Today, we build both native and web-based client applications using technologies like UWP, WPF, Electron and Angular, React and many other frameworks. Once again we’re experiencing the rise of another new approach to web-based client application development: WebAssembly. WebAssembly – or WASM – is a new, portable, size and load time efficient format, suitable for compilation to the web. I believe our industry has found its new silver bullet to bring strong typed languages to the web browser and write applications that run everywhere. In the past we saw all kinds of crazy frameworks and solutions popping up based on plugins. WASM opens the same set of possibilities as with plugins, but in a standardized way supported by all popular browsers. This means you can now create cross-platform UI frameworks that run natively in the browser and are written in C#, Java, Go or any other language. Microsoft is pursuing this concept with its Blazor project. Blazor is a framework for creating web pages that run C# code using .NET inside a web browser on any device or platform that supports the WASM web standard. People are even working on porting good old Silverlight code over to WebAssembly.

Of course, this is a great new way of doing things. However, one thing I have learned during the last few years in our industry is that there is nothing as volatile as client UI technology. Especially with the web having the next new shiny framework every six to nine months, and now probably even faster with WebAssembly. Will it become a big thing? Probably, but one of the things I would urge everyone to consider is to look at application development in a different way. We have a stable backend part and a very volatile frontend UI part. You are better off investing more in a good and robust, well performing back-end system, probably with a microservice architecture, running in containers on a cloud-managed Kubernetes cluster. The client side investment should be as low as possible. Use the flavor that makes developers happy at that moment, but keep in mind that you will probably need to rewrite it after a short period of time. My advice is also not to invest a penny in writing a client framework that would speed up your development or abstract other UI frameworks. It has proven to be a disinvestment. You’re better off coding the client in a straightforward manner, knowing you’ll discard it soon. This allows you to be more agile to adopt new client technologies and create a stable reliable and secure backbone that can serve any new frontend technology that may arise during the next few years.

Explosion of computing devices and generation of data

One final trend to touch on is the explosion of devices and data. It is unbelievable that the amount of data generated every day is over 2.5 quintillion bytes! The next few years more than 20-50 billion devices will be deployed to gather data, process data, and provide us with computing power. This is a game changer, because it provides us with such rich datasets that we can train our machine learning models better and better each day. Computers will be able to assist us in many more tasks than possible today. Machine learning, or AI as the industry calls it, will be incorporated into many things in our lives. From the smart doorbell to the smart office and the programs we use today, all will have AI capabilities that will make our lives easier. You can think of our planet becoming one gigantic computer, calculating and reasoning on our collective lives all the time. Again, this is a beautiful thing and scary at the same time. On the one hand, we can do great things to help mankind. On the other hand, we could also use it for malicious intent. As Satya Nadella put it at the Microsoft’s //build/ developer conference keynote: “All this computing power, and this collective computer we build together, will require us to think about what computers can do, but also about what computers should do. It is up to us and our governments to provide the legislation so we can ensure we will live in freedom and safeguard our privacy when we want to.”

One thing is clear to me if you look at all these changes in our industry. Things are changing faster than ever! The only constant I see is change. And you’d better embrace it, so you can create and shape the future now!

This article is part of our latest magazine; XPRT.#6 Download it here or get your free copy.