The idea

Recently, over here in continental Europe, there’s been a wee bit of a panic over the fact that ““we”” (sorry for the “politician speak”) as a part of the global society apparently are too dependent on the US and other big nations when it comes all things digital. To counter this, ““we”” must develop our own capabilities and thereby become less dependent.

Great idea, but what does this entail?

In this post, I’ll try and pick apart the landscape of compute, network and storage to figure out if this is even possible. The reason I’m asking the question here is because there is an awful lot of complication hidden beneath the idea of independence that i’d like to explore.

Dependency

What are dependencies then? The dependencies this story are about the form of “lock in” that occurs when you acquire a piece of digital technology. The acquisition makes it harder for you to have precise choice on what you see best fit as composition of solutions for a common compute problem you’re trying to solve.

Switching cost

The opinion of the composed solution you get served by the vendor of the technology brings forward a switching cost. The cost only appears if you want to solve the problem the way you see best fit. This switching cost is the “lock in” risk. Because now, if you want to change you have to invest more. But, do you really need to?

Lock in

This “lock in” materializes as follows. unique ideas are expressed, patented, encoded in solutions, potentially licensed and usage is constrained (by either companies or countries). Then a composition of many ideas is created and shipped as a product or service.

Compute, storage, networking. It sounds so simple.

We talk about compute, network and storage as if these we’re ancient relics, but this only has become more popular recently with the advent of cloud and commodity ubiquitous compute.

Von Neumann

To actually become wholly independent, and more significantly completely incompatible, we’d have to go back to the times of Von Neumann. He was the brilliant scientist that contributed a lot to the modern world of computing. Of course, he build his ideas on prior ideas going all the way back to Babbage and even before that Descartes, mind you.

For the thought experiment, let’s work from Von Neumanns time onwards, we’d have to develop our own ideas on “how to compute in a machine”. As you can see, this becomes a silly exercise really quick. So, we can already conclude that there is no getting away from that.

The hard parts

For the sake of continuing the thought experiment, I’m aiming to replace commodity compute, not some specialized form of compute for which EU players might be available in abundance.

When I deconstruct a server in my mind, I end up with these parts/layers for which we’ll evaluate how sovereign ““we”” are with respect to these parts.

The overlooked parts

We’ve now decomposed a server in relatively obvious parts, but of course this goes a few levels deeper. But a significant part, that is overlooked easily is “server management”. More and more pieces of hardware are getting tied to a cloud of some shape/form. This applies to servers, applications, devices, middleware, data platforms, development platforms, etc. This is the actual tricky part where a lot of proprietary technology is tied to the compute we use. This is also the software we cannot run the systems without in a responsible manner.

Carving out bigger moats

Within the playing field we see competing companies continuously trying to create competitive advantage, for example by carving out bigger moats around their products and offerings. Exclusively offering experiences on selected stages, using specific Apps, tools and platforms. This is where interoperability gets lost and one has to be really careful not to get “locked in” permanently.

Ecosystem and interoperability

As we can see, there is a vast ecosystem of things, all tightly interconnected to serve the purpose of providing common compute. This represents 50+ years of computing technology evolution. A state of technology, that despite being the field of strong competition also has built strategic trust among the players. The trust is built on the jointly created standards bodies and, as a form of leniency, offering of some patents under FRAND, the Fair, Reasonable and non-discriminatory use, to enable interoperability in the ecosystem.

The notion of this ecosystem is vastly important since otherwise we would not have ubiquitous compute, we would only have specialized compute.

One more thing - Open Source

The notion that we can replace just about anything with Open Source is also flawed. When we look for example to all major initiatives that are governed under CNCF, we see there is a huge influence (and hence continuity risk) in these projects. If we just look at these projects alone:

• Kubernetes
• Prometheus
• Envoy
• Helm
• Containerd

We can clearly see that these exist under strong influence of Google, Microsoft, Red Hat (IBM), VMWare and Apple. If US big tech gets sanctioned, these projects potentially will also suffer.

As stated before: ““There is no “getting away from this””

Then what?

A couple things come to mind:

1. Stop these BS discussions on “we need to become digitally sovereign”, which are useless and wasteful.
2. Let’s accept we live in a technology ecosystem, that both creates opportunity and dependency
3. Improve the conditions to thrive for European tech by providing cheap and abundant energy, reasonable taxes and incentives, facilitate a community that seeks growth and learning, make it rewarding to invest for individuals and companies.
4. Let’s focus on what we’re good at: “Building bridges”. This helps to connect communities, drive standards, create a thriving ecosystem and provide diplomacy where needed

Thanks for attending my TED Talk ;-)

Don’t forget to tell me what you think.