Back to blog

The attack was only 80 bytes long

Attack Was Only 80 Bytes Long

NetSenX Team11 min read
ndrincident-responseevidence

The most dangerous signal in your network may be smaller than a single sentence

At 3:17 a.m., a printer in the administration department sent a tiny packet of data to an external server. Eighty bytes. Less information than a short text message. The connection lasted a fraction of a second, was encrypted and did not lead to a domain appearing on any known threat list.

One minute later, the printer did it again. Then again. And again.

It was not downloading large files. It was not sending gigabytes of data. It did not attempt to access the financial server or execute software recognised as malicious. Every individual action appeared so insignificant that it would have been easy to ignore.

The problem was not the size of the packet. The problem was the rhythm.

Every sixty seconds, the device was informing an external system that it was still active and ready to receive instructions. This behaviour is known as beaconing, which means regularly sending short signals from a compromised device. It can be part of C2 communication, meaning Command and Control, the channel through which an attacker controls an infected computer, server, camera, printer or industrial device.

In this example, the attack was only eighty bytes long.

That is exactly why it could remain invisible.

Would your company’s security system recognise a threat revealed not by its content, but by the mathematical regularity of its behaviour?

The most effective attacks do not always look dangerous

For many years, security systems were designed primarily to recognise known patterns. When a file, address or fragment of communication matched a previously documented attack signature, an alert was generated.

Those controls are still necessary. The problem begins when an attacker uses a new tool variant, legitimate cloud infrastructure or encrypted communication that cannot easily be inspected.

The domain may appear legitimate. The certificate may be valid. The packet may be tiny. The connection may use a protocol relied upon every day by thousands of lawful applications.

In such situations, the threat is revealed not by what it is, but by how it behaves.

A compromised printer may contact an external server at perfectly regular intervals. An employee laptop may begin sending short packets a few seconds after every restart. An industrial camera may establish connections with a location it has never communicated with before.

Each event may be harmless in isolation. Their rhythm, history and relationships may tell a completely different story.

Do your security controls see individual events, or can they recognise the pattern developing between them?

The network remembers how devices behaved before

To recognise abnormal behaviour, an organisation must first understand what normal behaviour looks like.

A mature NDR system, meaning Network Detection and Response, should build an individual behavioural baseline for each device, account, process and service.

A printer normally communicates with a print server and downloads updates from its manufacturer. A camera sends video to a local recorder. The computer used by the finance team connects to banking and accounting systems during relatively predictable hours.

When one of these devices begins behaving differently, a signal appears.

That does not automatically mean an attack. The company may have introduced a new service, installed an update or changed suppliers. This is why a system should not react merely because something new has happened.

It should evaluate the context.

Did the new communication appear after a legitimate update? Are similar devices using the same connection? Does the amount of transferred data match the device’s function? Is the activity taking place during normal hours? Is the timing natural, or does it resemble an automated control signal?

Behavioural analysis does not ask only whether a device performed an allowed operation. It asks whether the current behaviour still makes sense for the role that device performs inside the organisation.

When was the last time someone in your company checked which systems a printer, camera, climate controller or forgotten warehouse computer was actually communicating with?

Eighty bytes may be completely harmless

A small, regular packet is not automatically evidence of an attack. Many legitimate systems send short control signals. Applications check whether servers are available, devices report their status and cloud services confirm that sessions remain active.

If every regular connection generated an alert, security teams would quickly be buried under thousands of warnings.

This is why the relationship between signals matters.

Has the device communicated with this destination before? Is the target server associated with a supplier used by the company? Is the timing almost perfectly regular? Did the connections begin after an unusual process or failed login? Are similar devices behaving differently? Did scanning activity or an attempt to reach another system appear at the same time?

Only by combining these factors can an organisation distinguish legitimate telemetry from emerging C2 communication.

The security world does not consist of only two states. Safe or dangerous. Attack or no attack.

Between them lies uncertainty, and that uncertainty must be measured honestly.

A system should be able to say that one signal is weak, while also showing that its significance increases because several other deviations appeared at the same time. It should also be able to admit that there is still not enough information to support a reliable conclusion.

Which system would you trust more? One that turns every unusual packet into a red alert, or one that explains why several apparently harmless events together create meaningful risk?

The biggest problem is not always a failure to detect

Security teams often receive enormous numbers of warnings. Some come from firewalls, others from endpoint protection, cloud services, email systems, servers and identity management tools.

The real signal may already be somewhere in the queue.

It may simply be buried beneath hundreds of false alarms.

A SOC, meaning Security Operations Center, does not become more effective merely because it receives more data. It needs fewer and better decisions supported by context and evidence.

When every new connection generates an alert, analysts stop treating warnings seriously. When a system alerts too rarely, it may miss a real threat. The difficult task is controlling the boundary between the two.

No honest security system can guarantee zero false positives and complete detection of every possible attack. It is, however, possible to control alert quality mathematically, define error boundaries under clearly stated assumptions and avoid pretending to be certain when the available evidence is insufficient.

Conformal prediction is a statistical method that can create verifiable uncertainty boundaries. In practice, it can help determine whether the false alarm rate remains under control instead of relying on a marketing claim that a solution is highly accurate.

A chief executive does not need to understand all the mathematics behind the method. Leadership should still expect a security provider to demonstrate the limits of its effectiveness and clearly identify the moment when there is not enough evidence to support a claim.

Why do we still accept promises from systems making security decisions when those promises cannot be independently verified by the customer?

One packet does not reveal the whole incident

The first small signal may be only the beginning.

A compromised device confirms that it is ready. It later receives an instruction to scan the network. The attacker then attempts to access another system, moves between devices and begins transferring information.

This progression is often described as an attack chain.

It may begin with reconnaissance, meaning the process of exploring the environment. The next stage may be initial access, when the attacker gains the first foothold. This may be followed by lateral movement, meaning movement between systems inside the organisation. A C2 channel may then be established, and the final stage may involve exfiltration, meaning the removal of data from the company.

Different tools may detect different stages. One system sees the scanning. Another notices an unusual login. A third records the external connection.

Who connects those events into one coherent story?

A company does not need five unrelated alerts. It needs one incident view showing which device was the starting point, which systems were reached, how the behaviour changed and where the complete path was leading.

A network map should reveal the relationships between the device, process, identity, domain, supplier and external destination. Attack-chain analysis should connect weak signals into a chronological sequence. An incident view should group related alerts so that an analyst does not need to reconstruct the story manually across dozens of screens.

After a serious incident, will your team see the full attack path, or spend several days comparing exports from different tools?

A good alert should answer why

The message “suspicious communication detected” is too vague to support a responsible decision.

An analyst, IT manager and board member should understand why the behaviour was considered risky.

An alert may explain that the device began communicating with a new external destination, packets appeared at nearly identical intervals, similar devices showed no comparable activity and the connections started outside normal operating hours. The risk assessment may increase further if network scanning or an attempt to access another system was observed beforehand.

The explanation should show which factors influenced the decision, how the current activity differs from the historical baseline and how much uncertainty remains. It should also allow the assessment process to be reproduced later.

Without this layer, artificial intelligence becomes another black box. The company receives a score, but not the knowledge needed to act.

Does your security system explain why an alert was created, or does it expect people to trust the colour displayed on a dashboard?

Not every suspicious packet should stop the company

Automatic blocking can be useful when a threat is unambiguous. In industrial, financial or production environments, however, an incorrect decision may cause an expensive interruption.

If a small packet is mistakenly classified as C2 communication, automatically disconnecting the device may stop a production line, disrupt a logistics system or disable a service needed by customers.

A responsible approach should therefore begin with a monitoring-first model. This means the system observes, classifies, explains and recommends before enforcement occurs. A person or a tool designed to enforce security policy can then make the decision to block the connection.

A firewall controls network traffic. EDR, meaning Endpoint Detection and Response, monitors endpoint devices such as computers and servers. SIEM, meaning Security Information and Event Management, collects and correlates logs. SOAR, meaning Security Orchestration, Automation and Response, automates agreed response procedures.

The NDR layer does not need to replace those systems. It should provide the information they often lack. What actually happened in network communication, why it matters and what evidence supports the recommendation.

Should security automatically disconnect a device after one uncertain signal, or first present the full context to the person responsible for the business consequences?

After the incident comes the question of evidence

Assume the small packet really was the beginning of an attack. The company detected the communication, isolated the device and stopped further activity.

That still does not close the matter.

The board, a customer, insurer, auditor or regulator may ask when the activity began, which devices were involved, whether data left the organisation, when the company became aware of the problem and what evidence supported the decision to respond.

A few screenshots will not be enough.

The organisation needs a decision trail showing the input data, the risk assessment, related signals, the event timeline, recommendations and actions taken. This record should be reproducible and suitable for use in a formal report.

NIS2 is the European Union directive on the security of network and information systems. It increases the responsibilities of many companies in relation to cyber risk management and incident handling. DORA is the European Union regulation on digital operational resilience in the financial sector. GDPR, known in Poland as RODO, regulates personal data protection and obligations related to data breaches.

No security system automatically creates full legal compliance with these regulations. It can, however, provide supporting evidence and workflow assistance that help an organisation demonstrate what it observed, how the event was assessed and which actions were taken.

Is your company prepared not only to detect an incident, but also to prove credibly what happened?

The board should not ask how many alerts were generated

The information that a security platform generated 427 alerts this month tells a chief executive almost nothing.

The board needs answers to different questions.

Was the company genuinely at risk? Which devices and business processes may have been affected? Did any information leave the organisation? Is the activity still continuing? How reliable is the assessment? What action is recommended? Could the incident affect customers, operations, reputation or regulatory obligations?

Technical data should be transformed into a business decision.

In the case of our eighty-byte signal, the board does not need to analyse every packet. It should see that a previously quiet device began communicating regularly with a new external destination, its behaviour differed from that of similar devices and later events may indicate a developing attack chain.

That is the difference between another dashboard and operational evidence.

Does your company’s security reporting help the board make decisions, or only demonstrate how much activity the tools have recorded?

The greatest threat may be almost invisible

Not every attack begins with a large data transfer, a red screen or a message saying that files have been encrypted. Sometimes it begins with a device almost everyone has forgotten and a packet so small that a person would never notice it.

An attacker does not always want to move quickly.

Sometimes the objective is to be regular, patient and boring.

That is why the future of threat detection will not depend only on more signatures. It will depend on the ability to understand behaviour, rhythm, relationships and change over time.

A strong security system should notice the small signal, but it should not panic because of it. It should connect the signal with context, control the risk of false alarms, explain its assessment and preserve evidence that allows the complete story to be reconstructed later.

The attack was only eighty bytes long.

The real question is whether your organisation would detect it before it grew into an incident that could no longer be hidden from customers or the board.

What is the greater cybersecurity problem today? A lack of data, too many alerts, or the inability to recognise the meaning of a small signal?

How many apparently insignificant connections are active inside your network right now, waiting for someone to notice their rhythm?

Sometimes the most important signal is not the loudest one. It is simply the most regular.

NetSenX is a European Network Detection and Response and Network Evidence platform that analyses network behaviour, detects C2 communication and beaconing, connects alerts into incidents and attack chains, and creates explainable evidence supporting security decisions and audit processes.

Further reading: what NetSenX detects and how each alert explains itself, NDR vs EDR vs SIEM for teams without a SOC and how NetSenX compares with other NDR vendors.


Written by the NetSenX Team at TriStiX S.L. Questions: info@netsenx.com.

NetSenX Team

TriStiX S.L.