What does it take to be a real data space?

-DATA SPACE / 2

When we started, we (I) honestly didn’t know what a real data space even is and I am still not entirely sure. But now I think I know more. Or at least enough to get into trouble. And my team hates when (mostly every day) I come with a new definition, idea, or just negate everything from the previous day.

THE PLAN

• Build a data space (at least Minimum Viable Data Space).
• Document all our failures and dead ends.
• Maybe, one day, declare: “This is it.”
• Or, more realistically: “This is way too big, complicated, undefined, whatsoever.” and with implementation of Pareto principle (20/80).

LET'S BEGIN AT THE BEGINNING

• What does a data space actually need to be called a data space?
• What is the bare minimum stack before we can slap the label data space ready on it?
• And how do we graduate from Minimum Viable Data space to FIWARE-ready and eventually Gaia-X-ready data space?

Because we’re building this with FIWAREBox (and because FIWARE looks like it’s going in the right direction), we decided to map the journey in three steps:

1. Minimum Viable Data space (MDS - bare minimum).
2. FIWARE-ready Data space (speaks NGSI-LD, eats SmartDataModels for breakfast plus some additional modules).
3. Gaia-X-ready Data space (hello European bureaucracy and federation).

Step 1: MV Data space – The Bare Minimum

At the most basic level, a data space is just a place where datasets are shared (published), discovered, and consumed. Nothing mystical. Nothing with AI.

The stack:

• Data Broker (Context Broker) – keeps our data alive instead of dead CSV files plus some sensor and simulated data can be added.
• Data Catalog – otherwise no one will ever find what we published.
• Identity & Access Management – because “open” doesn’t mean “open bar.”
• Authorization & Policy Enforcement – who gets the VIP pass, who doesn’t. And who controls this.
• Secure Data Exchange / Connectors – we don’t want our datasets sneaking out the back door. Seal everything!
• GUI / Developer Tools – a dashboard so people don’t think it’s just a black hole. And some tools to get their hands dirty.

With this stack, congratulations: we can already put “we run a MV data space” on our slides.

Step 2: FIWARE-Ready – The Open Standard Badge

To get into the FIWARE club, we need:

• MV Data space - with some modifications.
• NGSI-LD Compliant Broker – Orion-LD, Scorpio, Stellio, take your pick.
• Smart Data Models – because CSVs are sooo last decade.
• FIWARE Security Components – Keyrock, Wilma, AuthZForce (or something equally painful to configure).
• Standard Connectors – IDS / FIWARE Data space Connector (if it works this week).
• Monitoring & Observability – Prometheus + Grafana, so we can make nice dashboards when it fails.

Now we are officially not just a data space, but a FIWARE-ready data space. That’s progress.

Step 3: Gaia-X Ready – The Big League

Gaia-X is where things get serious. The moment you step in, it’s less about data and more about identity, compliance, and sovereignty. In short: paperwork, but digital.

We will need:

• Gaia-X Identity & Trust Framework – self-sovereign identity, verifiable credentials. Basically: prove you are who you say you are.
• Compliance Services – because “trust me” is not a policy.
• Data Sovereignty Services – making sure ourdataset doesn’t end up in places it shouldn’t.
• Federated Catalog Integration – because being alone is not Gaia-X.

Only then we can claim: “We are Gaia-X ready.” And then prepare to spend as much time explaining it as running it.

WORK IN PROGRESS

We’re still on this path. Half curious, half masochistic. Every week we learn something new, usually the hard way. But that’s fine.

Our promise: we’ll document it all — the good, the bad, and the why did we ever think this was a good idea?

Maybe at some point we’ll actually declare:

✔ - “This is it.”

Or, more likely:

X - “Let’s go back to CSV files, life was easier.”

Either way, we hope sharing this messy journey helps others who want to start small, stay sane, and maybe, just maybe, build a data space that actually works.

DATA SPACE SERIES CONTENT

So, What Exactly Is a Data space? 

DATA SPACE / 1

to be continued...

DATA SPACE / 3