Firewall rule base audit, UAE

Rule bases only ever grow. Adding a rule solves today. Removing one risks breaking something nobody remembers.

A firewall rule audit answers four questions about every rule: what does it permit, who asked for it, is it still used, and would anybody notice if it were removed. Most organisations can answer the first for all of them and the other three for almost none.

Firewall rule base audit for UAE organisations
  • Four questionsPer rule, and three are usually unanswered
  • Shadowed rulesRules that can never match anything
  • Any to anyThe finding that appears in almost every audit
  • CIS Control 12Network infrastructure management
What we examine

Seven findings that appear in almost every rule base we review.

Network infrastructure management is control 12 in the CIS Critical Security Controls at version 8.1, sitting alongside secure configuration at control 4 and network monitoring and defence at control 13. The rule base is where all three meet, and it is the artefact that most reliably records an organisation history rather than its intent.

Permissive rules that were meant to be temporary

An any to any rule, or a rule permitting a whole subnet to a whole subnet on all ports, added during a migration or an outage with every intention of tightening it afterwards. They are the single most common serious finding, they are usually near the top of the rule base where they shadow everything below, and nobody remembers adding them.

Shadowed rules that can never match

A rule that sits below a broader rule matching the same traffic will never be evaluated. It is not a security problem in itself, and it is strong evidence that the rule base has been edited by several people over several years without anybody looking at the whole. Shadowed rules also make every subsequent change harder to reason about.

Rules with no traffic against them

Usage data over a representative period tells you which rules are doing work and which are not. A rule with no hits over ninety days is a candidate for removal, subject to the obvious question of whether the period covered its use case. Seasonal, year-end and disaster recovery paths are the exceptions that make this a judgement rather than a script.

Rules with no owner and no justification

The question that closes most rules is not is it used, it is who asked for it and why. A rule with a change reference, a requester and a business reason can be reviewed. A rule with a comment field reading temp or a ticket number from a system that was replaced cannot, and it will survive every future review for exactly that reason.

Objects and groups that have drifted from their names

An address group called finance servers that now contains eleven objects, three of which are not finance servers and one of which is a whole subnet. Group drift is how a narrow rule silently becomes a broad one without the rule itself ever changing, and it is invisible unless somebody expands every object.

Configuration outside the rule base itself

Management access from anywhere, administrative accounts that are shared, logging disabled on rules that matter, firmware several versions behind, and high availability that has never been failed over. Secure configuration is control 4 in the CIS Controls, and a perfect rule base on a device anybody can log into is not a control.

Divergence between what the diagram says and what the device does

Almost every organisation has a network diagram that describes an intended architecture and a rule base that describes an accumulated one. The gap between them is where unexpected paths live, and reconciling the two is frequently the most useful output of the audit for anybody who has to design a change afterwards.

Why rule bases never shrink

Adding a rule has a clear benefit today. Removing one has an unclear risk tomorrow.

The asymmetry is structural, and it explains why rule bases grow monotonically in every organisation regardless of the quality of the team.

  • Adding a rule unblocks a project, resolves an incident or enables a supplier. The benefit is immediate, attributable and appreciated. It is the right thing to do at the time and it is almost never wrong.
  • Removing a rule has no visible benefit and a non-zero chance of breaking something nobody has documented. Nobody thanks the engineer who removed a rule successfully, and everybody remembers the one who removed the wrong one. The incentive runs one way.
  • The only thing that reverses the asymmetry is evidence. Usage data showing a rule has matched nothing over a representative period, plus a documented owner who confirms it is no longer needed, converts removal from a gamble into a decision. That evidence is what an audit produces.
  • This is also why a rule audit needs to end with a maintained process rather than a one-off cleanup. Without a review cycle and a requirement that new rules carry an owner and an expiry where appropriate, the rule base returns to its previous size within a few years.
Ask us to run a rule base review
How we approach it

Four things that turn a rule audit into a cleaner rule base.

Producing a list of findings from a rule base is largely mechanical. Getting rules actually removed is a different problem, and it is entirely about evidence and ownership rather than about analysis.

We use usage data, because opinion cannot remove a rule

Hit counts over a representative period turn removal from a gamble into a decision. Without them, every proposed removal is somebody asserting that a rule looks unnecessary, and that assertion never survives contact with the person responsible for whatever might break. The evidence is what makes the conversation short.

We trace ownership before recommending removal

A rule with a named owner can be confirmed or retired in one email. A rule with no owner requires either archaeology or a controlled disable and observe. Doing the tracing during the audit, rather than handing over a list of unattributed rules, is most of the difference between a report and a result.

We disable before we delete

The safe sequence for a rule with no recorded traffic and no identifiable owner is to disable it, observe for an agreed period, then remove it. That converts the worst case from an outage to a quick re-enable, and it makes the whole exercise something an operations team will agree to rather than resist.

We leave behind a process, not just a cleaner base

New rules requiring a named owner and a business justification, temporary rules carrying an expiry date, and an annual review with usage data. Without those three the rule base returns to its previous size within a few years, and the next audit produces the same report to a different audience.

Where this matters most

Six UAE situations where a rule audit pays for itself quickly.

The trigger is usually a migration, an audit finding, or an incident where somebody discovered a path nobody knew existed. All three are good reasons and the first is the cheapest.

An organisation migrating to a new firewall platform

Migrating a rule base is the single best opportunity to shrink it, because everything has to be examined anyway. Migrating it unchanged carries a decade of accumulated permissions onto a new platform and wastes the only occasion when removal is expected rather than resisted.

A regulated firm with a periodic review obligation

Several frameworks and payment obligations expect firewall rules to be reviewed at a defined frequency. Where such an obligation applies, the audit produces both the review and the evidence that it happened, with a documented decision against each rule rather than a signature confirming somebody looked.

A business in scope for payment card requirements

PCI DSS version 4.0.1 is the current version in the standards council document library, and network controls have always been central to it. Where card data is in scope, the segmentation the rule base implements is the control that determines how much of the estate is in scope, which makes rule accuracy commercially significant.

An operator with converged operational and corporate networks

Where production systems sit behind rules written over many years, the gap between the intended segmentation and the actual one is both a safety and a security question. Reconciling the documented design with the live rule base is frequently the most valuable single output for this profile.

A group that has grown by acquisition

Each acquisition brings a rule base written by a different team to different conventions, and integration typically adds permissive rules to make things work quickly. Auditing across the group produces a consistent picture and identifies the temporary integration paths that are now several years old.

An organisation that has just found an unexpected path

The usual prompt. Somebody discovers that one system can reach another that it should not, and the immediate question is what else is like that. A full rule audit is the only way to answer it, and the answer is almost always that there are several more, of which one or two matter.

Three positions

How UAE organisations manage their firewall rule bases.

The middle column is the norm. Changes go through a proper process, every rule was justified when it was added, and nothing has ever been removed.
New rules require justification
Audited and maintainedYes
Changes controlled, nothing removedYes
Unmanaged rule baseNo
Rules have an identified owner
Audited and maintainedYes
Changes controlled, nothing removedFor recent ones
Unmanaged rule baseNo
Unused rules identified
Audited and maintainedYes
Changes controlled, nothing removedNo
Unmanaged rule baseNo
Shadowed rules removed
Audited and maintainedYes
Changes controlled, nothing removedNo
Unmanaged rule baseNo
Object groups match their names
Audited and maintainedYes
Changes controlled, nothing removedDrifted
Unmanaged rule baseUnknown
Any to any rules known and scoped
Audited and maintainedYes
Changes controlled, nothing removedPresent, unnoticed
Unmanaged rule basePresent
Management access restricted
Audited and maintainedYes
Changes controlled, nothing removedUsually
Unmanaged rule baseNo
Documented design matches reality
Audited and maintainedYes
Changes controlled, nothing removedNo
Unmanaged rule baseNo design
Temporary rules expire
Audited and maintainedYes
Changes controlled, nothing removedNo
Unmanaged rule baseNo
Rule count trend
Audited and maintainedStable
Changes controlled, nothing removedRising
Unmanaged rule baseRising
The finding categories

Nine categories, and what we do with each.

This is how we classify findings, so the output is a decision list rather than an undifferentiated report. The recommendation column is a starting position, adjusted with the people who own the systems behind the rule.
FindingWhy it mattersDefault recommendation
Any to any, or any service permittedPermits far more than intended, and usually shadows rules below itScope to the actual requirement, urgently
Overly broad source or destinationA subnet where a host was neededNarrow to the specific systems, with the owner confirming
Shadowed ruleCan never match, and makes the base harder to reason aboutRemove, after confirming the shadowing rule is correct
No traffic over the review periodLikely obsolete, subject to seasonal exceptionsDisable, observe, then remove
No owner or justificationCannot be reviewed, so survives indefinitelyTrace the owner, or treat as a candidate for removal
Object or group driftA narrow rule that has silently become broadRebuild the group to match its name and purpose
Logging disabledNo record of what the rule permittedEnable, unless there is a documented volume reason
Management access too broadSecure configuration failure, independent of the rulesRestrict to a management network, with named accounts
Divergence from the documented designUnexpected paths that nobody designing a change would predictReconcile, and correct whichever is wrong
How an engagement runs

Five steps, and disabling before deleting is not negotiable.

Typically two to four weeks for the audit depending on the number of devices and the size of the rule bases, followed by a remediation window that suits your change process.
  1. 1

    Collect configuration and usage data

    Configuration exports from every in-scope device including branch and cloud firewalls, plus rule usage data over a representative period, ninety days as a reasonable minimum. Read-only access is normally sufficient, and a central management platform makes the collection considerably faster.

  2. 2

    Analyse the rule base against the finding categories

    Permissive rules, shadowed rules, rules with no traffic, rules with no owner or justification, object and group drift, logging gaps, and management access. Each finding is categorised so the output is a decision list rather than an undifferentiated report of everything that could be improved.

  3. 3

    Trace ownership and reconcile with the design

    Change records, ticket references and the people most likely to know, so that as many rules as possible arrive at the review with an owner attached. In parallel, the live rule base is compared with the documented network design, because the gap between them is where the surprising paths live.

  4. 4

    Agree the actions, then disable before deleting

    Urgent scoping for permissive rules, removal for shadowed ones, and disable and observe for rules with no traffic and no owner. The disable and observe step converts the worst case from an outage into a quick re-enable, which is what makes an operations team willing to proceed at pace.

  5. 5

    Leave a process that keeps the base clean

    A requirement that new rules carry a named owner and a business justification, an expiry date on anything described as temporary, and an annual review using usage data. Without those, the base returns to its previous size and the next audit produces the same findings.

Straight answers

What organisations ask about firewall rule audits.

Before the audit

Fifteen things that make the audit faster and better.

The first group is access, the second is context, and the third is what happens afterwards, which is what determines whether the rule base stays clean.

Access

  • Can we get a configuration export?
    Read-only access is usually sufficient.
  • Is rule usage data available?
    Hit counts over a representative period.
  • How long is the usage window?
    Ninety days is a reasonable minimum.
  • Are all devices in scope?
    Including branch and cloud firewalls.
  • Is there a central management platform?
    It makes the export far easier.

Context

  • Is there a network diagram?
    Even an out of date one helps.
  • Is there a change record for rules?
    That is where owners come from.
  • Are there seasonal or annual paths?
    They must not be removed on usage alone.
  • Any regulatory obligation in play?
    It may set a review frequency.
  • Which segments matter most?
    Prioritise the ones protecting real value.

Afterwards

  • Who approves a removal?
    Decide before the list arrives.
  • Is there a change window?
    Removals happen in batches.
  • Do new rules require an owner?
    The single most effective change.
  • Do temporary rules carry an expiry?
    Otherwise they are permanent.
  • When is the next review?
    Annually, or the base regrows.
Next step

Count the rules in your main firewall. Then ask how many were removed last year.

Tell us about firewall platforms, rule sets, traffic logs, application owners and change windows. We will confirm the requirements and proposed scope with you.