Glossary
Data flow mapping
Data flow mapping is the practice of recording and drawing the exchanges inside an IT estate: which application or machine sends what to which other, over which path and for what need. It combines a record of flows with the diagrams that show them, kept current over time.
Updated
What is data flow mapping?
Data flow mapping records the exchanges that cross an IT estate and draws them in a readable way. The mapping guide of ANSSI, the French cybersecurity agency, defines a flow as an exchange of information between a sender and a receiver, each of which can be an application service, an application or an actor (our translation).
Depending on who is asking, the same work is called application flow mapping, network flow mapping or simply a flow diagram: boxes for applications or zones, arrows for exchanges, a legend for the protocol or the kind of data.
The work pays off at specific moments. Before a migration, it says which interfaces will have to be reconnected. During an incident, it shows who is affected downstream of a stopped application. In a security review, it brings out exchanges between zones that have no reason to exist. Flow mapping is one part of mapping the IT estate, the part that connects the elements to one another.
Application flows, network flows, personal-data flows: what is the difference?
Three exercises share one name, and mixing them up wastes entire meetings.
Application flows describe what applications exchange in business terms: billing sends entries to accounting every night, the web store queries stock on every order. In ANSSI's guide they belong to the applications view, which describes software components, the services they offer and the data flows between them.
Network flows describe what travels between machines and between zones: a source address, a destination address, a protocol, a port. They belong to what the same guide calls the logical infrastructure view, which covers network segregation, address ranges, VLANs and filtering functions. A single application flow often becomes several network flows: sending the accounting entries involves a transfer server, a DNS lookup and an authentication against the directory.
Personal-data flows are a different exercise, carried out under the GDPR. The French data protection authority, the CNIL, in its method for mapping personal data processing, asks organizations to list flows with the origin and destination of the data, in particular to identify any transfers outside the European Union. That inventory starts from processing activities and feeds the record of processing the CNIL invites organizations to keep. It can draw on the technical map, which does not stand in for it.
What should you record for each flow?
ANSSI's guide proposes a short baseline for a flow: an identification and description, a sender, a receiver, and encryption. That baseline is enough for a first version.
Practice adds a few fields, depending on the level you work at. For an application flow: the kind of data exchanged, the direction, the frequency (on demand, hourly, nightly), the mechanism (API call, file drop, message queue) and an owner on each side. For a network flow: protocol, port, the zones crossed, and the justification, meaning the sentence that says what the flow is for.
Two more fields turn a record into a working tool. The source of each line says where the information came from: the project's flow matrix, a rule export, an interview with the application owner. The date says when it was last checked. A flow noted in an interview three years ago and a flow read last week in the firewall rules do not carry the same weight, and the diagram should let a reader tell them apart.
Names matter as much as columns. If the application is called Payroll on the diagram, HR-PAY in the CMDB and an IP address in the rules, the map cannot be matched against anything.
Where does the information come from, and how far can you trust each source?
Project flow matrices say what was requested at design time, with the justification. They usually date from go-live.
Firewall rules and, in the cloud, AWS and Azure security groups say what is permitted today. They are exact by construction, but they do not say why a rule exists or whether it is still used.
Logs and traffic observation show what actually flows. They reveal exchanges nobody had documented. An observed flow is not thereby a justified one, and a yearly job can fall outside the observation window.
Interviews with application owners provide what no file holds: the business meaning of an exchange, the job that sends it, what happens when it fails. They cost time and age quickly.
These sources describe two realities that should be kept apart: required flows, the ones applications need, and open flows, the ones the network lets through. An honest map shows both, and the gap between them. A required flow with no matching rule is an outage waiting for the next failover. An open rule with no required flow is an exposure nobody decided on.
How does a flow map differ from a flow matrix or a data flow diagram?
The flow matrix is the table: one row per flow, with source, destination, protocol, port, direction and justification. The flow map is the picture drawn from it and its upkeep over time: diagrams per zone or per application, groupings, and links to the rest of the map of the estate.
Each feeds the other. Nobody draws three hundred rows. The diagram groups servers into applications and subnets into zones, and points to the table for detail. In the other direction, the diagram often reveals the anomaly the table was hiding, such as an arrow from the office network zone straight into the database zone.
A data flow diagram, or DFD, is something else again. It is an analysis notation that shows how data moves between processes, data stores and external entities, with no concern for machines or ports. It helps design an application or analyze threats against it, whereas a flow matrix is what firewall rules are written from.
What are the common pitfalls?
The first pitfall is the single diagram. A map showing every flow of every application becomes unreadable beyond a few dozen elements. You need views by scope: one application and its neighbors, one zone and what enters it, one business process end to end.
Mixing levels does not help either. An arrow labeled ERP to data warehouse and an arrow labeled 10.2.3.4 to 10.8.1.12 over TCP 1521 cannot be read on the same drawing. Keep both levels, linked but separate.
The undated map misleads more quietly. A flow diagram that was exact on go-live day becomes wrong with nothing to signal it. Each view should carry the date it was last checked and say which sources it rests on.
Shared services are the classic omission. Directory, DNS, backup, monitoring and mail relay receive flows from almost the whole estate, and appear in no project's matrix.
Treating observed flows as intended flows costs more still. Rebuilding the map from logs alone amounts to declaring legitimate everything already in motion, including what should have been closed.
What Uncia brings, and what it does not do
Uncia reads, read-only, your flow matrices and the rules of your firewalls and clouds, through a connector or a dropped-in file, and reconciles them with the rest of the inventory. The generated network map shows zones, the firewalls between them and the flows crossing them. Uncia checks the matrix of required flows against the actual rules and returns a verdict per flow: covered, missing, inconsistent, unjustified. Each verdict says why, rule in hand, and every piece of information cites its source.
Uncia changes no rule and administers no firewall. It does not observe traffic and does not read logs: it compares what is written on one side with what is written on the other, and the result updates with each collection. It does not decide that a flow is legitimate; a person makes the call. It does not keep GDPR records of processing and does not map personal data. A flow that appears in neither your files nor your cloud accounts will not be on the map.
Sources
Frequently asked questions
What is the difference between a flow map and a flow matrix?
The flow matrix is the table that lists each flow with its source, destination, protocol, port and justification. The flow map is the picture drawn from it, per zone or per application, and its upkeep over time.
What is the difference between a data flow diagram and a flow matrix?
A data flow diagram (DFD) shows how data moves between processes, data stores and external entities, without machines or ports. A flow matrix lists specific network communications. The first helps design and analyze an application; the second is used to write and review firewall rules.
Is data flow mapping the same as GDPR data mapping?
No. The French data protection authority, the CNIL, asks organizations mapping their personal data processing to list flows with the origin and destination of the data, in particular to identify transfers outside the European Union. That work starts from processing activities. A technical flow map can feed it but does not stand in for it.
Where do you start with flow mapping?
With a small scope and sources you already have: project flow matrices and an export of firewall rules. Record a sender, a receiver and a description for each flow, then complete the picture with application owners.
Can you map flows from network logs alone?
You get the flows that exist, not the ones that are intended. That is useful for finding undocumented exchanges, but each observed flow then needs confirmation from the owner of the application concerned. A rare job can also fall outside the observation window.