Ask an IT leader to sketch how the company’s payroll system works, which databases feed it, which identity services control access to it, which network segments it crosses. The whiteboard usually stays mostly blank. Ask what else goes dark if one of those pieces fails, and the pause gets longer.
On an ordinary day, that blind spot rarely matters. It matters enormously the morning a ransomware crew detonates. In the first hour, responders need the blast radius. Which systems are encrypted? Which are only unreachable because something upstream failed? Which ones can the business not do without for another day? Teams that must rebuild that picture from tribal knowledge and stale spreadsheets lose hours they don’t have.
How that morning goes is largely decided months earlier. An organization that maps how its assets depend on one another, and keeps that map current, can sequence a recovery instead of improvising one.
Why the Inventory Isn’t Enough
Most organizations keep some record of what they own. For example, a configuration management database, an asset spreadsheet, a cloud console export. Those records tend to treat assets as isolated line items, and they fall behind as fast as engineers spin up new workloads. They’ll tell you a server exists and who owns it. They rarely tell you that three revenue-generating applications quietly authenticate through it.
Dependency mapping platforms take a different approach. They pull telemetry from the software, cloud and device integrations already in place and build a continuously refreshed model of how systems connect, upstream and downstream. The useful output is less a hardware inventory than a set of answers to operational questions. If this goes down, what follows it? Where does sensitive data move between zones that trust each other differently? What is the smallest set of systems the business needs back before anything else matters? Those answers feed risk modeling, incident response playbooks and the restoration order in a recovery plan.
A Map is Useless if the Data is Gone
Knowing the right restoration order does nothing if the backups the plan depends on were encrypted or deleted along with production. Attackers know this. Sophos surveyed 2,974 IT and security professionals whose organizations had been hit by ransomware, and 94% said the attackers tried to compromise their backups during the attack.
That’s where immutable storage earns its place, and why the label deserves scrutiny. Immutability usually comes from write-once storage or object-lock retention, which stop stored backups from being changed or erased until a retention period expires. But the protection depends on how it’s configured.
Amazon S3 Object Lock is a good example. Its compliance mode blocks deletion even by the account’s root user, while its governance mode lets users with a specific bypass permission delete protected data or shorten its retention. Attackers who reach a backup management console have also been documented modifying retention policies so that recovery points get erased. And an immutable copy of data that was already compromised is still compromised. Immutability keeps a backup intact. It doesn’t certify that the backup is clean.
Let Business Impact Set Backup Policy
Combining the two disciplines changes how backup policy gets written. Many teams apply one retention template across the estate to satisfy an auditor. Instead, they can use the dependency map to find three things: the services with the widest downstream reach, the single points of failure, and the paths an attacker could take from a low-value tier to a critical one. Those are the places for the longest retention, the strictest isolation and the most frequent restore testing.
The map also exposes weak spots in the backup environment itself. A backup console might sign in through the same directory as production, or service accounts might hold privileges on both sides. Either one gives an attacker with stolen production credentials a route to the recovery systems. With that visibility, teams can:
- Give backup administration its own identity boundary, separate from production directory services
- Segment backup networks so production hosts can’t reach them directly
- Apply zero trust controls (strong authentication, least privilege, per-session verification) to backup management
- Keep at least one copy offline or logically isolated in immutable storage. This matches CISA’s #StopRansomware guidance to maintain offline, encrypted backups and test them regularly.
Recovering in the Right Order
During an incident, the pairing pays off in sequencing. Before restoring anything, responders can confirm which recovery points predate the intrusion. They can bring identity services back first, so dependent applications have something to authenticate against. They can restore the services the business ranked highest instead of whatever comes back fastest. And they can avoid restoring into an environment the attacker still controls, which only restarts the infection. The result is a shorter mean time to recover (MTTR).
A current map also helps leadership with questions that now come with deadlines. Public companies, for example, must file a Form 8-K under Item 1.05 within four business days of determining that a cybersecurity incident is material. Scoping impact quickly is central to making that call.
What to Ask a Provider
Plenty of vendors can show you a dependency diagram, and plenty can sell you immutable storage. Far fewer can prove the two are connected. Ask how backup coverage is checked against the dependency map, and how often. Ask when they last ran a full restore test of a critical service, and how long it took. Ask whether they have rebuilt a production environment after a real ransomware attack, not a tabletop exercise.
The Opening for MSPs and MSSPs
For managed service providers, this is a chance to move beyond monitoring and alerting into something clients will pay a premium for: the ability to bring a business back. The providers that stand out will combine dependency intelligence, properly configured immutable backups and a tested restoration record into one resilience service. They won’t stop at detecting attacks. They’ll be able to show they can restore entire environments when defenses fail.