What classic antivirus does

Traditional antivirus checks files against a database of signatures: fingerprints of known malware. When a file matches a signature, it is blocked or moved to quarantine. Modern products have added basic behavior checks (heuristics) and cloud-based detection, but the principle stays the same: recognize something known and block it before it runs.

This is an important layer of protection, and it still stops a large amount of common malware that arrives by email, in downloads and on USB devices. The question is what happens when the attack does not look like a known malicious file.

Where antivirus is not enough

Many attacks today are not based on a single file that can be identified. An attacker who obtained a password (credential theft) signs in like a regular user. Another attacker uses legitimate tools that already exist on the computer (living off the land), such as PowerShell or remote management tools, so as not to leave suspicious files behind (fileless). New malware changes faster than signatures are updated, and is sometimes built specifically for one organization (targeted attack). An employee who installs legitimate remote control software at the request of someone who called and claimed to be from support (social engineering) will not trigger any antivirus alert either.

In cases like these the antivirus sees no problem, because as far as it is concerned nothing unusual happened. The organization discovers the attack only when the files are already encrypted, or when someone from outside reports that data leaked (data breach). At that point it is no longer a technical fault: work stops, you need to find out what data was exposed, and sometimes also report it to the Privacy Protection Authority and inform customers.

What EDR adds

EDR (Endpoint Detection and Response) focuses on the behavior of the computer. A component installed on every computer and server (agent) continuously collects information about what is happening: which processes are running, who started them, which files they access, where they connect on the network and what changes they make to the system. The information is sent to a central system that analyzes it and looks for attack patterns (indicators of attack).

This makes it possible to detect, for example, an Office document that launches PowerShell, a process that starts encrypting many files in a short time, or an attempt to pull passwords from memory (credential dumping). Each of these actions can look innocent on its own, and the context is what gives it away.

Another advantage is the full picture. When there is an alert, EDR shows the chain (attack chain): how the incident started, what happened at each stage and which other computers are involved. That is the difference between “a suspicious file was blocked” and understanding what the attacker tried to do, and whether they are still on the network.

Response: what happens when there is an alert

The second part of the name, Response, matters just as much. EDR systems let you act immediately and remotely: isolate a computer from the network (isolation) so it can only communicate with the management system, stop a process, delete a file, and in some products also roll back changes malware made (rollback). Quickly isolating one computer can keep an incident from spreading to the whole organization.

It is worth deciding in advance which actions will run automatically (automated response) and which require approval. Automatically isolating a workstation is usually a simple decision, but isolating a critical server during working hours may stop the business, so that decision needs to be planned.

The information collected is also used after the incident. It helps understand what happened (forensics), since when the attacker was inside (dwell time) and which systems were affected, and that is exactly the information needed when deciding whether there is a duty to report (breach notification) and what to tell customers.

XDR, MDR and SOC: the terms around it

  • XDR: an extension of EDR that connects information from more sources, such as email, identities, cloud and network, to detect attacks that move between systems.
  • MDR: a managed service in which a professional team monitors the EDR, investigates alerts and responds on behalf of the organization.
  • SOC: a security monitoring center, internal or external, that receives events from various systems, usually through a SIEM, and handles them.

For a small or mid-sized business, the practical combination is EDR with a managed monitoring service. Larger organizations add a SIEM and a SOC that bring together information from all systems, including the firewall, email and the cloud.

What to check when choosing

  • Coverage: Windows and Mac workstations, servers, and Linux if needed.
  • Protection against being disabled (tamper protection): an attacker who gains privileges will first try to turn off the protection. The product needs to make that hard.
  • Response capabilities: isolation, stopping processes and remote action.
  • Data retention: how long the data is kept for investigation after the fact.
  • Management and follow-up: who gets the alerts and how handling them is set up.
  • Impact on the computers: load on older workstations and compatibility with specialized software.

It is worth starting with a phased rollout: a small group of computers and servers, a period of following alerts and compatibility, and only then a full rollout. During this period the necessary exclusions are also defined, so that they stay narrow and documented.

EDR is one layer out of several

Even a good EDR does not replace the basics. Computers and servers that do not get security updates (patch management), accounts without MFA, employees with administrator rights (local admin) and a flat network in which every computer sees every other one leave the attacker too many ways in. EDR will detect some of them, but it is much easier to stop an attack that hits an obstacle at every stage.

In the same way, EDR does not replace backup. If ransomware managed to encrypt a server before it was stopped, what gets the business back to work is an isolated backup (immutable backup) and a restore that was tested in advance. The right picture is one of layers (defense in depth): updates, permissions, network separation, EDR, monitoring and backup, with each one covering what the others miss.

Where the IT company fits in

EDR touches every computer in the organization, so it has to fit into day-to-day management: installation on every new device, controlled exclusions for specialized software, handling alerts and coordination with whoever manages the servers and the network. When the IT company and the security function work separately, gaps appear: computers without the agent installed, overly broad exclusions added to solve a fault, or an alert each side thought the other was handling.

The right way is an IT company that understands information security in depth, and also knows what the business is required to do under the Privacy Protection Regulations (Data Security), 2017 and the National Cyber Directorate guidance. On the day of an incident, how fast the response is depends on whether whoever gets the alert knows the network and can act immediately.

Summary

  • Antivirus detects known threats. EDR detects suspicious behavior and lets you respond.
  • The value of EDR depends on who monitors and responds to the alerts.
  • For small and mid-sized businesses, EDR with managed monitoring (MDR) is a practical starting point.
  • Make sure the agent is installed on every computer and server, and that there is a clear process for handling alerts.