What a firewall does today

The firewall is the crossing point between the business network and the internet, and often between parts of the internal network as well. In the past its job was mainly to open and close ports. Today, a next-generation firewall (NGFW) identifies applications (application control), checks traffic against signatures of known attacks (IPS), filters browsing (web filter), manages VPN connections and serves as the control point between network zones.

So the choice of firewall determines what the business can see, what it can block and how employees connect from outside. A firewall that does not fit the business, or a good firewall that is badly configured, leaves it exposed even when it looks protected. And when it goes down, in most offices work stops too: no internet, no access to cloud systems and no connection to the branches.

Start with the needs, not the model

Before comparing vendors and models, it is worth answering a few questions about the business itself:

  • How many users and devices there are today, and how many are expected in the coming years.
  • What the speed of the internet lines is, and whether there is, or will be, a backup line (dual WAN).
  • Whether there are branches that need to be connected to each other (site-to-site VPN).
  • How many employees work remotely, and which systems they need access to.
  • Which servers or services are open to the internet, if any.
  • Whether there are regulatory or customer requirements for logging, monitoring and network separation.

The answers determine the size class of the device, the licensing needed, and whether two devices are needed in a high availability setup. At this stage it is worth involving both whoever manages the IT and whoever is responsible for information security, because the decision affects the work of both for years.

Performance: the number on the product page is not your number

Vendors publish several throughput figures for each model. The highest number usually refers to traffic without advanced inspection. When you turn on intrusion prevention (IPS), application identification, malware inspection (anti-malware) and especially decryption of encrypted traffic, actual throughput drops, sometimes significantly.

When evaluating a model, look at the throughput figures with all the protections you plan to turn on, and leave room for growth. A device that runs at the edge of its capacity on installation day will cause slowdowns before long, and the fix chosen then is usually to turn protections off. That is how a firewall bought to protect the business gradually turns into an expensive router.

Capabilities worth checking

  • Application and user identification: rules by application and by user or group from Active Directory, instead of by IP address only.
  • IPS and malware filtering: with regular signature updates from the vendor.
  • Web filtering: blocking dangerous categories and new or suspicious sites.
  • Support for VLANs and rules between zones: so the firewall also serves to separate the internal network (segmentation).
  • Multiple lines (multi-WAN): automatic switchover to a backup line (failover) when the main line goes down.
  • High availability: two devices that back each other up, for businesses where a firewall failure stops work.
  • Log forwarding: the ability to send events to a central system, a SIEM or a SOC service, and keep them for a period that allows investigation.

Remote access: VPN and ZTNA

A VPN connection lets employees connect to the business network from home or on the road. It is an essential capability, and also one of the points attackers look for most. VPN gateways are exposed to the internet by definition, and over the years serious vulnerabilities have been found in them at various vendors, some of which were actively exploited before an update was even available (zero-day).

A few basic principles: two-factor authentication (MFA) on every connection, fast updates to the firewall’s firmware when a vulnerability is published, and limiting each user’s access to only the systems they need (least privilege). The ZTNA approach extends this principle: instead of connecting the user to the whole network, they get access to a specific application only, after an identity check and sometimes also a check of the device’s state (device posture). It is worth checking whether the firewall you are evaluating supports this or integrates with such a solution.

The firewall’s own management interface should not be open to the internet. That sounds obvious, and it is still one of the common mistakes.

Licensing, updates and long-term support

In most commercial firewalls, the advanced capabilities depend on a subscription: IPS signatures, web filtering, malware protection and vendor support. When the subscription expires, the device keeps passing traffic, but the protections stop updating. It is important to understand in advance what is included, what needs to be renewed and what happens at the end of the term.

It is also worth checking the model’s lifecycle: until when the vendor provides security updates. A device that is out of support (end of life) is a risk, because new vulnerabilities found in it will not be fixed. Its replacement should be planned in advance, not a reaction to a failure.

Management, logs and who is responsible

A good firewall without ongoing management loses its value quickly. Temporary rules that stay open, postponed updates and alerts nobody reads are the common state in businesses where the device was installed and forgotten.

Proper management includes firmware updates, a periodic rule review, configuration backup, tracking of events and documentation of every change (change management). This is where the IT company’s role comes in: whoever manages the firewall needs to know current attack methods, the National Cyber Directorate guidance and the requirements of the Privacy Protection Regulations (Data Security), 2017. The firewall is usually one of the first components examined after a security incident, and its logs and settings determine whether it will be possible to understand what happened and explain it to the Privacy Protection Authority, the insurance company and customers.

Common mistakes

  • Choosing by device price alone: without calculating the subscriptions over several years and the work needed to manage it.
  • Leaving default settings: overly broad rules (any-any rules), an admin password that was never changed and open services nobody needs.
  • Unrestricted outbound traffic to any destination: every computer can reach any address on any port, so malware that got in can call out (command and control) without interference.
  • Internal services exposed to the internet: remote desktop (RDP) or a management interface opened “for a moment” to help a supplier, and left open.
  • No configuration backup: when the device fails, all the rules are rebuilt from memory.

Summary: questions to ask before buying

  1. What the business needs today and in a few years.
  2. What the model’s throughput is with all protections turned on.
  3. Which subscriptions are required and what happens when they end.
  4. How employees connect remotely, and with what authentication (MFA).
  5. Until when the model receives security updates (end of life).
  6. Who manages the device, updates it and reads the alerts.