Information Security Interview Questions

What follows is a list of questions for use in vetting candidates for positions in Information Security. Many of the questions are designed to get the candidate to think, and to articulate that thought process in a scenario where preparation was not possible. Observing these types of responses is often as important as the actual answers.

I’ve mixed technical questions with those that are more theory and opinion-based, and they are also mixed in terms of difficulty. A number of trick questions are included, but the goal there is to expose glaring technical weakness, not to be cute. I also include with each question a few words on expected responses.

–

Where do you get your security news from?

Here I’m looking to see how in tune they are with the security community. Answers I’m looking for include syndication feeds for solid sites like liquidmatrix, packetstorm, rootsecure, secguru, astalavista, whitedust, internet storm center, etc. The exact sources don’t really matter. What does matter is that he doesn’t respond with, “I go to the CNET website.”, or, “Steve Gibson’s home page”. It’s these types of answers that will tell you he’s likely not on top of things.

If you had to both encrypt and compress data during transmission, which would you do first, and why?

If they don’t know the answer immediately it’s ok. The key is how they react. Do they panic, or do they enjoy the challenge and think through it? I was asked this question during an interview at Cisco. I told the interviewer that I didn’t know the answer but that I needed just a few seconds to figure it out. I thought out loud and within 10 seconds gave him my answer: “Compress then encrypt. If you encrypt first you’ll have nothing but random data to work with, which will destroy any potential benefit from compression.”

What’s the difference between HTTP and HTML?

Obviously the answer is that one is the networking/application protocol and the other is the markup language, but again–the main thing you’re looking for is for him not to panic.

How does HTTP handle state?

It doesn’t, of course. Not natively. Good answers are things like “cookies”, but the best answer is that cookies are a hack to make up for the fact that HTTP doesn’t do it itself.

What exactly is Cross Site Scripting?

You’d be amazed at how many security people don’t know even the basics of this immensely important topic. We’re looking for them to say anything regarding an attacker getting a victim to run script content (usually Javascript) within their browser.

What’s the difference between stored and reflected XSS?

Stored is on a static page or pulled from a database and displayed to the user directly. Reflected comes from the user in the form of a request (usually constructed by an attacker), and then gets run in the victim’s browser when the results are returned from the site.

What are the common defenses against XSS?

Input Validation/Output Sanitization, with focus on the latter.

What’s the difference between symmetric and public-key cryptography

Standard stuff here–single key vs. two keys, etc, etc.

In public-key cryptography you have a public and a private key, and you often perform both encryption and signing functions. Which key is used for which function?

You encrypt with the other person’s public key, and you sign with your own private. If they confuse the two, don’t put them in charge of your PKI project.

What kind of network do you have at home?

Good answers here are anything that shows you he’s a computer/technology/security enthusiast and not just someone looking for a paycheck. So if he’s got multiple systems running multiple operating systems you’re probably in good shape. What you don’t want to hear is, “I get enough computers when I’m at work..” I’ve yet to meet a serious security guy who doesn’t have a considerable home network.

What is Cross-Site Request Forgery?

Not knowing this is more forgivable than not knowing what XSS is, but only for junior positions. Desired answer: when an attacker gets a victim’s browser to make requests, ideally with their credentials included, without their knowing. A solid example of this is when an IMG tag points to a URL associated with an action, e.g. http://foo.com/logout/. A victim just loading that page could potentially get logged out from foo.com, and their browser would have made the action, not them (since browsers load all IMG tags automatically).

How does one defend against CSRF?

Nonces required by the server for each page or each request is an accepted, albeit not foolproof, method. Again, we’re looking for recognition and basic understanding here–not a full, expert level dissertation on the subject. Adjust expectations according to the position you’re hiring for.

What port does ping work over?

A trick question, to be sure, but an important one. If he starts throwing out port numbers you may want to immediately move to the next candidate. Hint: ICMP is a layer 3 protocol (it doesn’t work over a port) A good variation of this question is to ask whether ping uses TCP or UDP. An answer of either is a fail, as those are layer 4 protocols.

How exactly does traceroute/tracert work at the protocol level?

This is a fairly technical question but it’s an important concept to understand. It’s not natively a “security” question really, but it shows you whether or not they like to understand how things work, which is crucial for an Infosec professional. If they get it right you can lighten up and offer extra credit for the difference between Linux and Windows versions.

The key point people usually miss is that each packet that’s sent out doesn’t go to a different place. Many people think that it first sends a packet to the first hop, gets a time. Then it sends a packet to the second hop, gets a time, and keeps going until it gets done. That’s incorrect. It actually keeps sending packets to the final destination; the only change is the TTL that’s used. The extra credit is the fact that Windows uses ICMP by default while Linux uses UDP.

If you were to start a job as head engineer or CSO at a Fortune 500 company due to the previous guy being fired for incompetence, what would your priorities be? [Imagine you start on day one with no knowledge of the environment]

We don’t need a list here; we’re looking for the basics. Where is the important data? Who interacts with it? Network diagrams. Visibility touch points. Ingress and egress filtering. Previous vulnerability assessments. What’s being logged an audited? Etc. The key is to see that they could quickly prioritize, in just a few seconds, what would be the most important things to learn in an unknown situation.

As a corporate Information Security professional, what’s more important to focus on: threats or vulnerabilities?

This one is opinion-based, and we all have opinions. Focus on the quality of the argument put forth rather than whether or not they they chose the same as you, necessarily. My answer to this is that vulnerabilities should usually be the main focus since we in the corporate world usually have little control over the threats.

Another way to take that, however, is to say that the threats (in terms of vectors) will always remain the same, and that the vulnerabilities we are fixing are only the known ones. Therefore we should be applying defense-in-depth based on threat modeling in addition to just keeping ourselves up to date.

Both are true, of course; the key is to hear what they have to say on the matter.

Describe the last program or script that you wrote. What problem did it solve?

All we want to see here is if the color drains from the guy’s face. If he panics then we not only know he’s not a programmer (not necessarily bad), but that he’s afraid of programming (bad). I know it’s controversial, but I think that any high-level security guy needs some programming skills. They don’t need to be a God at it, but they need to understand the concepts and at least be able to muddle through some scripting when required.

What are Linux’s strengths and weaknesses vs. Windows?

Look for biases. Does he absolutely hate Windows and refuse to work with it? This is a sign of an immature hobbyist who will cause you problems in the future. Is he a Windows fanboy who hates Linux with a passion? If so just thank him for his time and show him out. Linux is everywhere in the security world.

What’s the difference between a threat, vulnerability, and a risk?

As weak as the CISSP is as a security certification it does teach some good concepts. Knowing basics like risk, vulnerability, threat, exposure, etc. (and being able to differentiate them) is important for a security professional. Ask as many of these as you’d like, but keep in mind that there are a few differing schools on this. Just look for solid answers that are self-consistent.

Cryptographically speaking, what is the main method of building a shared secret over a public medium?

Diffie-Hellman. And if they get that right you can follow-up with the next one.

What’s the difference between Diffie-Hellman and RSA?

Diffie-Hellman is a key-exchange protocol, and RSA is an encryption/signing protocol. If they get that far, make sure they can elaborate on the actual difference, which is that one requeres you to have key material beforehand (RSA), while the other does not (DH). Blank stares are undesirable.

What kind of attack is a standard Diffie-Hellman exchange vulnerable to?

Man-in-the-middle, as neither side is authenticated.

What’s the goal of information security within an organization?

This is a big one. What I look for is one of two approaches; the first is the über-lockdown approach, i.e. “To control access to information as much as possible, sir!” While admirable, this again shows a bit of immaturity. Not really in a bad way, just not quite what I’m looking for. A much better answer in my view is something along the lines of, “To help the organization succeed. ”This type of response shows that the individual understands that business is there to make money, and that we are there to help them do that. It is this sort of perspective that I think represents the highest level of security understanding—-a realization that security is there for the company and not the other way around.

Are open-source projects more or less secure than proprietary ones?

The answer to this question is often very telling about a given candidate. It shows 1) whether or not they know what they’re talking about in terms of development, and 2) it really illustrates the maturity of the individual (a common theme among my questions). My main goal here is to get them to show me pros and cons for each. If I just get the “many eyes” regurgitation then I’ll know he’s read Slashdot and not much else. And if I just get the “people in China can put anything in the kernel” routine then I’ll know he’s not so good at looking at the complete picture.

The ideal answer involves the size of the project, how many developers are working on it (and what their backgrounds are), and most importantly — quality control. In short, there’s no way to tell the quality of a project simply by knowing that it’s either open-source or proprietary. There are many examples of horribly insecure applications that came from both camps.

What’s the difference between encoding, encryption, and hashing?

Encoding is designed to protect the integrity of data as it crosses networks and systems, i.e. to keep its original message upon arriving, and it isn’t primarily a security function. It is easily reversible because the system for encoding is almost necessarily and by definition in wide use. Encryption is designed purely for confidentiality and is reversible only if you have the appropriate key/keys. With hashing the operation is one-way (non-reversible), and the output is of a fixed length that is usually much smaller than the input.

Who do you look up to within the field of Information Security? Why?

A standard question type. All we’re looking for here is to see if they pay attention to the industry leaders, and to possibly glean some more insight into how they approach security. If they name a bunch of hackers/criminals that’ll tell you one thing, and if they name a few of the pioneers that’ll say another. If they don’t know anyone in Security, well…consider closely what position you’re hiring them for. Hopefully it’s a junior position.

Advanced

Ok, now for some more advanced questions:

1. If I’m on my laptop, here inside my company, and I have just plugged in my network cable. How many packets must leave my NIC in order to complete a traceroute to twitter.com?
   
   The key here is that they need to factor in all layers: Ethernet, IP, DNS, ICMP/UDP, etc. And they need to consider round-trip times. What you’re looking for is a realization that this is the way to approach it, and an attempt to knock it out. A bad answer is the look of WTF on the fact of the interviewee.



2. How would you build the ultimate botnet?
   
   Answers here can vary widely; you want to see them cover the basics: encryption, DNS rotation, the use of common protocols, obscuring the heartbeat, the mechanism for providing updates, etc. Again, poor answers are things like, “I don’t make them; I stop them.”

Bonus: Scenario Role-Play

For special situations you may want to do the ultimate interview question. This is a role-played scenario, where the candidate is a consultant and you control the environment. I had one of these during an interview and it was quite valuable.

So you tell them, for example, that they’ve been called in to help a client who’s received a call from their ISP stating that one or more computers on their network have been compromised. And it’s their job to fix it. They are now at the client site and are free to talk to you as the client (interviewing them), or to ask you as the controller of the environment, e.g. “I sniff the external connection using tcpdump on port 80. Do I see any connections to IP 4.2.2.2?” And you can then say yes or no, etc.

From there they continue to troubleshooting/investigating until they solve the problem or you discontinue the exercise due to frustration or pity.

–

Feel free to contact me if you have any comments on the questions, or if you have an ideas for additions.

Shoichi Nakagawa faces questions about wobbly performance at G-7 meet as government announces economy shrank at 12.7% rate in fourth quarter.

It wasn't the biggest economic contraction in more than three decades that piqued the attention of Japan's media on Monday. It was the apparent drunken attempt of the country's finance minister, Shoichi Nakagawa, to respond to questions at a press conference in Rome following the end of a Group of Seven meeting that fueled chatter in Tokyo.

Eyes drooping and red-faced, Nakagawa slurred his replies and fumbled his answers to reporters in Italy's capital on Saturday. Greeting him back at work in Tokyo on Monday was unequivocal evidence that Japan's economy is the worse for wear. In the three months to the end of the year, the world's second-biggest economy shrank by an annualized rate of 12.7% as exports collapsed. (See "Japan Hits The Skids" and "Asia's Economic Dragons Wheezing")

In a sign of how consumers have shunned made-in-Japan products, Toyota Motor (nyse: TM - news - people ), the nation's leading manufacturer, expects to lose close to $5 billion this year compared with a profit of almost $7 billion a year earlier. (See "Moody's Puts The Boot Into Toyota") Toyota and the rest of the Japanese economy may be in for an even bumpier ride going forward.

"Everything indicates that Q1 will be worse. It's a very severe economic adjustment," said Glenn Maguire, chief Asia-Pacific economist for Societe Generale in Hong Kong. Making matters worse for Japan is the yen. As investors take refuge in the Japanese currency, its value against the dollar and other currencies has gained some 40% in recent months. That means "further retrenchment in capacity and labor," added Maguire, who predicts that average real GDP contraction in 2009 will be between 5% and 8%.

In Rome, Nakagawa, along with other G-7 finance ministers and central bank heads, agreed that the state of the world economy was dire, describing it in a statement as "severe." The group also committed itself to "act together using the full range of policy tools to support growth and employment and strengthen the financial sector.

Back in Tokyo, Nakagawa reportedly blamed his condition on having imbibed too much cough medicine. Speaking on television, one former prime minister, Yoshiro Hori, admonished Nakagawa for his Rome performance.

Nakagawa's boss, Prime Minister Taro Aso, is struggling to deliver on the G-7 commitment. Stimulus measures have been bogged down by political wrangling both within the ruling Liberal Democratic Party and with the opposition Democratic Party Of Japan, which controls the upper chamber of Japan's parliament. With an approval rating dipping below 10% and a national election looming, the beleaguered prime minister, who promised that Japan would be the first industrialized nation to emerge from recession, may have to leave it to his successor to fix the economy.

The reality, reckons Societe Generale's Maguire, is that Japan will be "first in and last out," as a quarterly recovery is unlikely until the second half of 2010. With little hope to sustain it, reaching for the bottle may be the only way for some Japanese to dull the economic pain.

For Merck and Schering-Plough, it is the $5 billion question: Does Zetia prevent heart attacks and strokes or not? It has been 13 months since Forbes reported the results of the first study that asked that question, setting off a chain reaction of scientific and governmental inquiry that has dramatically pushed down sales of Zetia and its sister drug, Vytorin, which together generated some $5 billion in annual sales last year.

The study, called ENHANCE, showed no benefit to the arteries when Zetia was added to the off-patent Zocor. Vytorin is a combo pill of Zetia and Zocor, and is sold by Merck (nyse: MRK - news - people ) and Schering in a joint venture. Shares of both companies have plummeted, and Merck will try to shore up investor confidence at a meeting with analysts tomorrow. Now a new analysis from a study funded by the National Institutes of Health seems to show Zetia works on the same measure of artery disease it flunked in ENHANCE.

Has Zetia been exonerated? No. The new study, SANDS, is the first new evidence that ENHANCE may have been wrong. In the words of J.P. Morgan analyst Chriss Schott, it's "finally a positive data point." It will provide some comfort to doctors prescribing the drugs as national guidelines already suggest: as options for people who can't tolerate more well-proven cholesterol-lowering drugs because of side effects. But it is unlikely to head off the next iceberg that Wall Street is worried about: that insurers will make getting the Vytorin combo pill more expensive during the next year. And it emphasizes how Merck and Schering have failed to do the studies needed to defend their $5 billion-a-year (sales) cholesterol franchise.

William James Howard of Washington Hospital Center, the lead investigator of SANDS, firmly believes his results are a strong counterargument to ENHANCE, which he calls "unfortunately not a sound scientific study." He is an expert in cholesterol who does paid speaking for many drug makers. He says his study "has given some comfort that Zetia is a rational choice." Both ENHANCE and SANDS measured how much thicker patients' arteries got, using ultrasound cameras to look at arteries in the neck. ENHANCE tested the Vytorin combo (Zetia plus Zocor) against Zocor alone in patients with a genetic disease that causes super-high cholesterol, and the addition of Zetia made no difference. SANDS wasn't designed to evaluate the question of what Zetia adds to a cardiovascular regimen. It showed that among a group of Native Americans with diabetes, healthier arteries result from more aggressive treatment of both high cholesterol and blood pressure. The initial results were published in the Journal of the American Medical Association in April.

In an effort to tease out the effect of Zetia, researchers looked at a subgroup of patients who were supposed to lower their cholesterol to very low levels. Sixty-nine people got Zetia on top of more proven cholesterol drugs and 154 didn't. So long as they lowered their LDL, or bad cholesterol, by the same amount, their arteries did equally well, Howard says. This analysis was published in the current Journal of the American College of Cardiology.

Howard argues that the arteries of the patients in ENHANCE had simply been cleared out by years of treatment with cholesterol drugs, called statins, and so it was impossible to show a difference. Cincinnati cardiologist Evan Stein, one of the coauthors of the ENHANCE paper and a big proponent of the too-thin argument, reiterated it in an editorial that appears alongside SANDS in the cardiology journal.

The argument that walls were too thin got a boost from a recently presented analysis by John Kastelein of University Hospital Center in the Netherlands, the lead investigator of ENHANCE. He looked at a number of studies that showed in patients treated with genetically super-high cholesterol who'd been treated, it was getting difficult to show a difference in artery thickness. The patients had been treated so much with statins that it was difficult to show a difference. But Kastelein also presented another analysis showing that in ENHANCE, differences in cholesterol level based on Zocor resulted in differences in artery thickness. So adding Zetia to Zocor in that study still should have resulted in thinner arteries.