bastion

Vulnerability Assessment and Security Testing

Build this skill

Distinguish a vulnerability scan from a penetration test, state what written permission must cover, and prioritise scan findings by exposure and impact.

Forty-seven findings and a question from Vikram

On the first Monday of December, Neha Kulkarni opens a report from Sentry Desk listing 47 weaknesses across Tulsi Health's 14 billing laptops, the customer app's server and the store routers. Two are marked critical, nine high, and the rest medium or low. The same morning Vikram Deshmukh asks whether they should hire someone to "try to break in" before the festive sale. Both are forms of vulnerability assessment: finding the weaknesses in your own systems and ranking them before an attacker finds them. The report came from an automatic scan; Vikram is asking about a penetration test. They sound alike, but they answer different questions, and both depend on one thing first: the owner's written permission.

Vulnerability scanning: an inspector with a checklist

A vulnerability is a weakness that someone could misuse to cause harm, such as software with a known flaw or a door left unlocked. Picture a building inspector visiting the Kothrud store with a long printed checklist. They walk through every room and tick boxes: fire extinguisher in date, exit sign lit, lock fitted, window latch firm. They write down what fails, but they never start a fire to see whether the extinguisher works, and they never try to force the lock.

A vulnerability scanner is a program that does this inspection for computers. It contacts each machine on a list and compares what it finds against a catalogue of thousands of known weaknesses. It looks for software versions with published flaws, missing patches (the small fixes that software makers release to repair flaws), default passwords that were never changed, risky settings, and open ports. A port is a numbered doorway on a computer through which a program accepts connections from elsewhere, a little like the numbered counters at a post office; a counter that is open but nobody needs is one more way in. Each finding is given a severity, often on the CVSS scale from 0 to 10, where 9 and above counts as critical.

Scanning is wide, fast, cheap and repeatable, so it is done regularly, every month and after every big change. A scan that logs in to each machine with an account Neha provides, an authenticated scan, can look at installed software from the inside and gives more accurate results than one that only knocks on the outside. Scanners have clear limits, though. They only know weaknesses that are already catalogued. They produce false positives, findings that look bad but are wrong, for instance flagging an old version number when the fix was in fact applied. And they cannot tell that the printer matters far less than the customer database.

Penetration testing: a locksmith with a signed letter

Now imagine Vikram hires a skilled locksmith, gives him a signed letter, and asks him to try to get into the Kothrud store after closing time. The locksmith does not just list loose latches. He tries them, and if the window latch gives, he climbs through to see whether the back office door is also weak and whether the safe is within reach. His report says exactly how he got in, how far he got, and what would have stopped him.

A penetration test, or pentest, is that visit for computer systems: an authorised, simulated attack by a skilled person, a penetration tester or ethical hacker, to discover how far a real attacker could get. Its main strength is chaining, combining small weaknesses into a serious path. A tester might find that the store's guest Wi-Fi is not properly separated from the staff network, and that a billing laptop runs an old MedLedger version; a scanner lists both as medium, but together they let someone sitting in the shop reach customer orders. Testers usually stop at proving the access, taking a screenshot rather than copying real data, and they avoid anything that would harm the business. Depending on how much they are told beforehand, a test may be black box (no inside knowledge, like an outsider), white box (full details and diagrams) or grey box (somewhere between). Because a pentest takes days of an expert's time, it is done less often, typically yearly or before a big launch.

Vulnerability scan TULSI HEALTH CHECKLIST ✓Lock fitted ✓Alarm on !Old lock model !Latch loose ✓Exit sign lit …200 more Wide, automatic, regular Lists the weak spots it can see Never tries to get in Penetration test TULSI HEALTH TOOLS PERMISSION Scope: Kothrud store systems Dates: 3–5 Dec Stop if asked Signed: Vikram D. ✓ Deep, by a person, rare Proves what an attacker could reach Only with signed permission

The inspector notes the loose latch; the locksmith shows that the loose latch leads to the back office. Only the locksmith tries the door, and only because he carries a signed letter.

Vulnerability scanAutomatic programEvery system, every monthFinds known weaknessesSome false positivesHours, low costPenetration testSkilled personChosen systems, yearlyChains weaknesses togetherProves real accessDays, higher cost

A scan answers 'what weaknesses do we have?'; a pentest answers 'what could an attacker actually do with them?'. Most organisations need both.

Sort the statements below into the activity they describe. Some are about breadth and routine, others about depth and human judgment; decide before you check.

Permission first, every time

The locksmith and the burglar may use the same skills; the signed letter is the only thing that separates them. An authorised security assessment is one done with the written permission of someone who has the right to grant it, for systems they actually own or control. In India, accessing a computer system without its owner's permission is an offence under the Information Technology Act, 2000, and good intentions are not a defence. The same scan or test that is professional work with permission is a crime without it.

A proper permission document, often called the rules of engagement, settles the details before anyone starts. Some systems may belong to someone else: MedLedger is Rathi Softech's software and the customer app runs on a cloud provider's computers, so their rules and agreement may be needed too.

ScopeExactly which systemse.g. Kothrud store onlyOut of scopeWhat must not be touchede.g. payment gatewayTime windowDates and hourse.g. 3–5 Dec, after 22:00MethodsWhat is allowedNo real data copiedContactsWho to call, day or nightand when to stopSigned bySomeone with authoritye.g. Vikram Deshmukh

If any of these is missing or unclear, the test does not start. Testers stop immediately when told to, even mid-test.

The same rule applies to you as a student. Suppose Riya notices that changing one number in her college portal's web address shows another student's marks. The right move is to stop there, not to look at more records "to be sure", and to tell the college IT office calmly what she saw. Some companies publish a bug bounty or disclosure policy that invites outsiders to test within written rules; that published policy is the permission, and its limits must be followed exactly.

Fixing in the right order

Fixing a weakness is called remediation, and nobody can fix 47 things on one Monday. The order matters more than the speed. Severity is only the starting point. Neha also asks how exposed each weakness is (reachable from the internet, or only from inside a store), how valuable the system is (2.4 lakh customer records, or a printer), whether criminals are known to be using that flaw right now, and how hard the fix is. When a fix must wait, for example until Rathi Softech releases an update, a temporary protection such as closing the affected port keeps the risk down meanwhile.

Scanfind weaknessesPrioritiserisk, not just scoreFixpatch or protectVerifyrescan the fixEvery month and afterevery change

A finding is closed only when a fresh scan confirms the fix. Skipping the verify step is how 'fixed' weaknesses survive for years.

Score six findings from Neha's December report for likelihood and impact below, then rank them and compare with the reference reasoning. Notice that one critical-looking finding may be a false positive, and that a modest score on a busy internet-facing server can outrank a scary one on a printer.

Follow your curiosity.

↑ ↓ to explore · Enter to open · Escape to close