Industry executives and experts share their predictions for 2023.  Read them in this 15th annual VMblog.com series exclusive.

The Metaverse Grows Up

By members of the Mirantis team

Prediction: 2023-5 actually will be “the years of the Metaverse,” with practical business applications beyond virtual worlds hitting mainstream.

A couple of my colleagues attended the Open Source in Finance Forum in New York, last week. There, they met the “Metaverse Guy” for a major merchant bank, who’s leading a team of about 20 developers, building simulation and collaboration apps for virtual and augmented reality.

My colleagues had been “the Metaverse Person” for one or another first-tier high-tech company about twelve years ago, during the heyday of Second Life. So, they were quick to point out the differences between today’s metaverse and that legacy model, effectively dependent on a single provider.

First – yes, there’s a lack of standards for platforms and devices. But nobody is waiting for standards. The “metaverse guy for a major merchant bank” said his team was betting on game engines like Unreal and Unity, for now – both of which permit building multi-user 3D environments in isolation from whatever aspiring virtual world providers decide to do. Meanwhile, he said his team is abstracting away from devices already. They’re just assuming things will need to work gracefully behind many different device contexts, and building that abstraction layer so they can move faster with applications.

Consumer applications aside, he suggests, the real juice in metaverse will be in straight-up business applications, and that ‘immersive’ metaverse experiences will start informing many parts of the workday in many domains: from individual desktops to collaborative environments.

Some examples: already, products are coming available that let a desktop worker manifest multiple large, high resolution displays in a virtual space. Coders, traders, and other big “screen real-estate consumers” will adopt these, particularly as AR devices get smaller and lighter and the remaining small issues with eye-tracking and head-tracking get ironed out. An obvious extension of this to collaboration gets you quickly to “virtual NOCs” – collaborative, multi-screen environments where many folks can meet, look at screens together, and solve problems.

We’ll also see the manifestation of “digital twinning,” in which physical objects such as industrial machinery, scientific instruments, cloud infrastructure, etc., are fitted with sensors or emit data that makes them both visible and manageable from within a 3D environment. Already, digital twinning is used to enable simulating impacts of changes before making them in the real world; 3D visualization will make digital twinning even more tangible and useful. 

Nations and government entities are now getting involved. For example, the tiny island nation of Tuvalu is creating a digital twin of its land even as it disappears beneath the ocean, trying to preserve its culture, if not its physical location. The European Space Agency is creating a digital twin of the earth that includes in situ measurements and observations and uses artificial intelligence to help predict both human and natural patterns. The US government has a military metaverse called Combined Augmented Reality Battlespace Operation Network (CARBON). Police departments in Alabama are using headsets to train their officers in how to work with the public. Europol has released a report called “Policing in the metaverse: what law enforcement needs to know.” Look for these kinds of things to become mainstream, starting this year.

Platform Abstraction: Not Serverless, Not PaaS, but a Secret, Third Thing

Prediction: 2023 and beyond will see rapidly-improving Kubernetes developer experience, reduced operations workloads, and faster application delivery. This will be due to new platform-abstraction frameworks that – without imposing the limitations of serverless or the overheads and new complexity of PaaS – will simply automate operations for a wide range of common application design patterns and operational footprints.

Despite all the talk about “shifting left,” developers still aren’t being served well in cloud-native environments. They already need to know a ton of stuff to write the business logic parts of applications – the parts that actually deliver value. And we’re telling them they now also need to learn a stack of new things: like cloud-native design patterns, and how to containerize components, and build and maintain CI/CD, and realize production application topologies with YAML configs and other Kubernetes constructs.

Serverless and PaaS work against this trend, sure. But serverless is too simplistic for many use-cases, and PaaS adds its own learning-curves, overheads, and constraints.

What we’re starting to see, though, is more like light-duty, specialized platform abstractions for different common application types, like the open source project, Lagoon. These frameworks really do let developers completely avoid the largest parts of the ops burden. The framework understands the standard components – web servers, proxy servers, databases, etc. – knows how they go together, and knows how to operationalize this design pattern on Kubernetes. So developers can write their business logic and push it with a very simple configuration, and get a fully-operationalized, self-scaling, self-healing, best practices-compliant result, with application-appropriate observability and other good stuff built in.

Cloud operators, meanwhile, can look at these platforms and understand, modify, and extend them, because they’re light and Kubernetes-native – unlike a PaaS where the platform is so heavy it creates a whole new domain operators need to master.

This is quickly going to expand beyond the simplest design patterns and application models: probably, makers will pick the lowest-hanging fruit first and productize those solutions: building the integrations around components in descending order of popularity (graph databases, etc.).

AI-assisted coding: more and better

Prediction: Existing models for AI-enhanced coding and conversational AI will fuse into more powerful tools to improve code quality and accelerate development.

Introduction of “AI pair programming” services like CodePilot have obviously gotten a lot of attention, as well as raising red flags of all kinds around IP provenance, open source and other licensing compliance, and potentially security. Meanwhile, coders we know have started using these services, and are finding them useful.

Coders have also started playing with the popular large language models from OpenAI and others – things like ChatGPT. These are currently less dependably useful – for example, ChatGPT can explain classic algorithms pretty well, but is challenged by simple math word problems like “if five machines can make five widgets in five hours, how long will it take ten machines to make ten widgets?” ChatGPT has been seen (your mileage may vary) to confidently answer “ten hours,” and then confidently (and wrongly) explain why.

But this is going to improve like gravity – fast, and accelerating. It’s just a matter of training on more and more curated data. And, so we think we’re very quickly going to see a fusion of these two branches of AI technology to optimally solve problems in the coding sphere with expert-level proficiency.

We’ll see applications like automatic code optimization at the conceptual level – versus the kind of optimizations compilers do. Optimization that really depends on modeling the problem versus looking only at the code. We’ll see code validation in arbitrary languages – something that’s now very difficult to do, even when working in languages specifically designed to simplify validation. We’ll see improved application security. And we’ll probably see more “security from the bottom” stuff where the LLM is put to work analyzing, for example, the SBOMs of component containers and specifying how to rebuild them to serve the same dependencies but contain fewer, or no, CVEs. It’s going to be scary, maybe, but increasingly great.

Clouds Everywhere

Prediction: Compute expectations are going to change with more work being performed locally on the desktop, on mobile devices, and on an increasingly-light and mobile “edge,” taking advantage of technologies like WASM and increasingly stripped-down container runtimes and new orchestrators and platform abstractions.

Over the past couple of years, we’ve all bought Raspberry Pis – the smarter of us before prices started skyrocketing – and built little Kubernetes clusters on them. They work – well enough, now, that in some cases, they’re inducing repatriation of certain kinds of software development and testing work, and even some production work, off public clouds and onto a new kind of network edge that lives on tiny little devices, long-running virtual machines on laptops, and in other interstitial places.

Meanwhile, technologies like WASM are providing interesting alternative ways to design applications, to exploit available compute wherever it lives, and make business logic smaller and smaller – as well as faster to build. In a sense, WASM makes the ‘write once, run anywhere’ and ‘abstract away the platform’ premises of Java and virtualization and containers much more real; you write to a stack-based virtual machine that can live anywhere.

This is going to result in the creation of brand-new platform abstractions, extensions of existing platform models, and probably fusions of both, to make ‘write once, deploy and scale anywhere’ much more seamless, reducing burdens on devs to learn new platform abstractions and keep track of breaking changes on fast-evolving popular standards.

##

ABOUT THE AUTHORS

Shaun O’Meara is global field CTO at Mirantis and has worked with customers designing and building enterprise IT infrastructure for 20 years.

Nick Chase, director of technical marketing for Mirantis, is a former software developer, Oracle instructor, and author of hundreds of tutorials and more than a dozen books on various programming topics.

John Jainschigg is director, open source initiatives at Mirantis, and is a cloud engineer, software developer, content/product marketer, and technology journalist.