Data Gravity, Egress Fees and Lock-in: It’s time to move on

The concept of data storage is as old as the computer itself — and frankly, it shows. Most organizations today are multi-cloud by design. And despite cloud-first strategies, almost all still have some level of on-prem infrastructure — even if it’s just for user endpoints or internal systems. But as environments multiply — and as each cloud or provider introduces its own tools and formats — silos become inevitable.

Multi-cloud and hyperscalers don’t just encourage fragmentation. In many ways, they’re built on it. Every provider — not just hyperscalers — offers a buffet of storage services: block, file, object, SQL, NoSQL… you name it. But behind that choice is a single goal: Create lock-in.

Every time you adopt a new cloud service, you inherit a new way to store, access, and protect your data — often incompatible with the rest of your environment. So you end up with multiple copies of the same data — each governed by a different set of security and protection rules, often imposed by the providers themselves. It’s not just duplication. It’s fragmentation you can’t control. This isn’t just costly — it’s unsustainable. Sooner or later, you lose visibility and control over your data.

Storage systems were built for a different world

Let’s look back. Historically, data storage evolved to match the need of the moment. At first, it was built for a single machine — the local hard drive. Then came centralized systems in the data center, like SAN and NAS. Eventually, with the rise of the internet, we moved to object storage — capable of serving data to millions of devices across the globe. But while the front end evolved, the back end didn’t. The core design remained the same: all data was still concentrated in a single location — even as our applications became globally distributed.

Let’s rewind for a moment. When cloud started making its way into the enterprise, everyone looked to the big providers for innovation — especially in managed services. And yes, they delivered on many fronts. But if you take a closer look at storage, not much really changed. Take AWS and S3, for example. It felt like a breakthrough at the time. But behind the scenes? It’s still built on the same old model:

• Your data lives in a single datacenter.
• Need a copy? You have to replicate it to another region.
• Want to access it from elsewhere? You point your application to a specific DC endpoint.
• Interoperability? Still missing.

In short: we dressed up old limitations in new interfaces.That may have worked then. But hybrid and multi-cloud changed the game. You now need to create and access the same data across clouds, on-prem, and the edge — and do it without adding cost, latency, or complexity.

It’s time to rethink your storage systems

To move forward, we have to stop thinking in terms of storage systems, and start designing data infrastructure. A data infrastructure is not a storage solution. It’s a platform — an abstraction layer — that spans your on-prem, cloud, and edge environments. It lets you create, access, and manage data anywhere, with consistency. With the right data infrastructure, you can:

• Apply one security and protection policy across environments
• Eliminate duplicate silos and redundant copies
• Maintain full visibility and governance
• Focus on value, not location
• And most importantly: regain control

When you decouple your data from the underlying hardware, you can quickly adapt to new business changes and evolve your IT strategy based on your organization’s needs. When you're no longer stuck with a specific cloud provider — held back by lock-in or the data gravity that cripples any attempt to adopt new technologies — you're free to innovate. That independence becomes a competitive advantage.

How do you get there?

There’s no magic product. But there are foundational principles that make data infrastructure possible:

• Metadata management Keep metadata separate from data — and manage it independently.
• Abstraction and virtualization Or software-defined, if you prefer the term. The goal is to decouple logic from hardware.
• Embedded security Encryption is just the beginning. Think holistic, end-to-end security models.
• Distributed architecture Keep your infrastructure mobile and adaptable to shifting business needs.
• Single data domain One global view, one namespace. Keep it simple and usable — complexity kills visibility.

These are the building blocks that let your data move with your business — not against it.

Takeaways

Data storage is a legacy mindset. If your organization is aiming for a sustainable hybrid or multi-cloud strategy, it's time to evolve. Why? Because:

• It increases control over data
• It gives you freedom to evolve the infrastructure underneath — including migrations, expansions, or replacing obsolete tech

Data gravity still exists, but when your workloads and data are spread intelligently, the impact is reduced. Every change is less painful, and lock-in from egress fees becomes a minor hurdle — not a hard stop.

Found this relevant?I share weekly thoughts on cloud, data strategy, and long-term IT resilience.Follow Control-Alt-Delete to stay ahead of the curve.