Skip to content

Digital Resilience — and the Ghost Redundancy Nobody Mentions

  • by

Systems Architecture & Resilience

Digital Resilience – and the Ghost Redundancy Nobody Mentions

Why redundant systems fail exactly when they are needed most, and how to identify the hidden points of “Shared Fate.”

Elias worked with a specialized tuning hammer that had been in his family for three generations. He was a piano tuner, a man who lived in the narrow, vibrating space between “almost” and “perfect.” On a Tuesday morning in a drafty community center, he was wrestling with an old Bechstein that hadn’t seen a felt-tip pen, let alone a tuner, in a decade.

He wasn’t just tightening wires; he was managing roughly of kinetic tension distributed across a cast-iron plate. He knew that if the plate had a hairline fracture, every turn of his hammer was a gamble. You can have the finest, most redundant set of strings in the world, but if the frame they are anchored to fails, the piano doesn’t just go out of tune-it ceases to be an instrument. It becomes a very heavy box of junk.

Elias understood something that most IT directors and clinic managers forget: backup systems are often just different strings anchored to the same cracked frame.

The Illusion of the Armored Desktop

Marta sat in her office three hundred miles away from the nearest IT hub, participating in what her clinic called a “resiliency drill.” She was proud of her setup. Her desk featured a traditional landline, a sleek VOIP headset, and a tablet with a dedicated 5G SIM card.

If the local fiber optic line was severed by a wayward backhoe, she could still reach her patients. If the power dipped, her uninterruptible power supply would hum into life for at least . She felt insulated, armored against the chaotic unpredictability of the physical world.

📞

Landline

Traditional Analog

🌐

5G Mobile

Secondary Gateway

🔋

UPS Power

90-Minute Buffer

The drill began at . The administrator “killed” the primary internet connection. Marta didn’t blink. She switched her tablet to the cellular network and picked up the office phone. The dial tone was a comforting, steady drone.

She was ready to begin her afternoon of telemedicine follow-ups. She had sixteen patients scheduled, many of them living in remote areas where the air was thick with the kind of industrial particulates that make a simple breath feel like a chore.

The Critical Disconnect

She reached for the patient manifest. She clicked the icon for the scheduling software. The screen didn’t load. It timed out.

She tried the secondary login portal, the one the vendors promised was a “fail-safe” entry point. It, too, hung in a white void of inactivity. Marta had the phone, she had the data connection, and she had the will to work.

But the register-the underlying database that held the phone numbers, the clinical histories, the specific allergen profiles of the children she was supposed to call-was hosted on the same AWS instance as the primary portal.

She sat there, listening to the rhythmic, mocking hum of her backup power supply, realizing she didn’t even know who she was supposed to be calling. The redundancy was a ghost.

Understanding “Shared Fate”

In the world of system architecture, this is often called “Shared Fate.” It is the silent killer of continuity. We build elaborate labyrinths of secondary and tertiary access points, but we often forget to check if those paths eventually converge at a single, brittle point of failure. It is like having a spare key to a car that has no engine.

Portal A

Portal B

AWS Database

SHARED FATE

The “Tightly Coupled” trap: Multiple access points connected to a single centralized brain.

To understand how this actually works, you have to look at the “Source of Truth” in a data environment. Most modern healthcare platforms use a distributed front end-the parts you see and click-but they often rely on a single, centralized database cluster to ensure data integrity.

This is a deliberate choice. You don’t want two different versions of a patient’s medical record floating around. If Marta updates a dosage on the backup system, it needs to reflect on the main system instantly. To achieve this, the systems are “tightly coupled.”

The unintended consequence is that any “heart attack” at the database level kills every interface simultaneously. Real redundancy requires “asynchronous replication,” where a second, independent copy of the data exists in a different physical location, ready to take over the moment the primary “brain” stops firing. But that is expensive, complex, and rarely checked during a sunny-day drill.

The Skin and the Splinter

I spent yesterday afternoon hunched over a desk lamp with a pair of tweezers and a magnifying glass, trying to extract a cedar splinter from the meat of my thumb. I had successfully ignored it for , thinking it would just work its way out.

Instead, the skin had begun to knit itself over the entry point, sealing the irritant inside a small, angry red dome. It reminded me that the “fix” we apply to a problem is often just a layer of skin over a deeper disruption. If you don’t address the core-the splinter, the cracked frame, the single database-you are just decorating a disaster.

Social Impact & Continuity

In the realm of social healthcare initiatives, the stakes for this kind of continuity are significantly higher. When you look at an organization like Projeto Brasil Sem Alergia, which has managed over 714,000 care encounters over the last , you aren’t just looking at a business; you’re looking at a massive web of human dependency.

Legacy of Care

714,000+

Encounters Managed

For over , maintaining data integrity across has been vital for patient safety.

They operate 10 in-person clinics across major hubs like Rio de Janeiro, São Paulo, and Curitiba. For many of the families they serve, these clinics are the first and only line of defense against chronic respiratory distress or severe skin reactions.

The project has expanded its reach through things like the Plano Diamante, which utilizes telemedicine to provide follow-up care for patients who live far from the physical clinics in Baixada Fluminense or other urban centers. For these patients, the digital link isn’t a luxury; it’s the bridge to their immunotherapy.

If that bridge is built on a “shared fate” architecture, the patient in a rural village is the one who pays the price when the “redundant” system fails.

Continuity isn’t a feature you buy; it’s a discipline you practice. It’s the uncomfortable process of asking, “What happens if the thing we trust to catch us also falls?”

True Resilience in Practice

True resilience in healthcare means recognizing that a virtual follow-up is only as strong as the accessibility of the data behind it. If a patient needs to

agendar consulta alergista,

they aren’t just looking for a slot on a calendar; they are looking for the assurance that their history, their tests, and their previous reactions are known and accessible, regardless of whether the primary server in a distant data center is having a bad day.

We often congratulate ourselves on having “more than one way to join a meeting.” We have the link in our email, the link in our calendar, and the app on our phone. But if the meeting ID itself is generated by a service that is currently down, all those entry points are just different ways to arrive at a locked door.

The Integrity of the Frame

I think back to Elias and his Bechstein piano. He told me once that the hardest part of his job wasn’t the tuning-it was telling an owner that their beautiful, polished instrument was structurally unsound. They would point to the shiny mahogany and the pristine ivory keys, refusing to believe that the iron heart of the thing was broken.

We do the same with our digital systems. We look at the beautiful UI and the “99.9% Uptime” stickers on the marketing materials, and we forget to look at the frame.

Routine uptime is a liar. It tells you that because things are working now, they will work when the pressure increases. But uptime doesn’t measure independence. It doesn’t tell you if your fallback is just a mirror of your failure.

To build something that actually lasts-whether it’s a 30-year social project or a simple telemedicine link-you have to be willing to hunt for the splinter. You have to look past the “two ways to connect” and ask if both of those ways are holding hands under the table, waiting to trip over the same obstacle.

A backup phone is just a plastic weight when the register holding the voice on the other end has vanished.

The next time you are told that a system is resilient because it has “multiple access points,” think of Marta in her quiet office, listening to the dial tone of a working phone that has no one to call. Think of the 714,000 encounters that rely on more than just a working internet connection. They rely on the integrity of the frame.

It is easy to be redundant when everything is fine. The real test of a system-and perhaps a person-is what remains when the “shared fate” finally catches up with the primary plan. We don’t need more front doors. We need a register that doesn’t disappear when the lights go out.