The last time I checked, the majority of networking and security professionals were still human.
We all know that the problem with humans is that they sometimes exhibit certain behaviors that can lead to trouble – if that wasn’t the case we’d probably all be out of a job! One such behavior is obsession.
Obsession can be defined as an idea or thought that continually preoccupies or intrudes on a person’s mind. I’ve worked with a number of clients who have had an obsession that may, as bizarrely as it seems, have had a negative impact on their information security program.
The obsession I speak of is the thought of someone “breaking in” to their network from the outside.
You’re probably thinking to yourself, how on earth can being obsessed with protecting your network from external threats have a negative impact on your security? If anything it’s probably the only reason you’d want a penetration test in the first place! I’ll admit, you’re correct about that, but allow me to explain.
Every organization has a finite security budget. How they use that budget is up to them, and this is where the aforementioned obsession can play its part. If I’m a network administrator with a limited security budget and all I think about is keeping people out of my network, my shopping list will likely consist of edge firewalls, web-application firewalls, IDS/IPS and a sprinkling of penetration testing.
If I’m a pen tester working on behalf of that network administrator I’ll scan the network and see a limited number of open ports thanks to the firewall, trigger the IPS, have my SQL injection attempts dropped by the WAF and generally won’t be able to get very far. Then my time will be up, I’ll write a nice report about how secure the network is and move on. Six or twelve months later, I’ll do exactly the same test, find exactly the same things and move on again. This is the problem. It might not sound like a problem, but trust me, it is. Once we’ve gotten to this point, we’ve lost sight of the reason for doing the pen test in the first place.
The test is designed to be a simulation of an attack conducted by a malicious hacker with eyes only for the client. If a hacker is unable to break into the network from the outside, chances are they won’t wait around for a few months and try exactly the same approach all over again. Malicious hackers are some of the most creative people on the planet. If we really want to do as they do, we need to give our testing a creativity injection. It’s our responsibility as security professionals to do this, and encourage our clients to let us do it.
Here’s the thing, because both pen testers and clients have obsessed over how hackers breaking into stuff for so long, we’ve actually gotten a lot better at stopping them from doing so. That’s not to say that there will never be a stray firewall rule that gives away a little too much skin, or a hastily written piece of code that doesn’t validate input properly, but generally speaking “breaking in” is no longer the path of least resistance at many organizations – and malicious hackers know it. Instead “breaking out” of a network is the new route of choice.
While everyone has been busy fortifying defenses on the way in to the network, traffic on the way out is seldom subject to such scrutiny – making it a very attractive proposition to an attacker. Of course, the attacker still has to get themselves into position behind the firewall to exploit this – but how? And how can we simulate it in a penetration test?
What the Pen Tester sees
The Whole Picture
There is no surer way of getting on the other side of the firewall than to head to your clients office and plugging directly into their network. This isn’t a new idea by any means, but it’s something that’s regularly overlooked in favor of external or remote testing. The main reason for this of course is the cost. Putting up a tester for a few nights in a hotel and paying travel expenses can put additional strain on the security budget. However, doing so is a hugely valuable exercise for the client. I’ve tested networks from the outside that have shown little room for enumeration, let alone exploitation. But once I headed on-site and came at those networks from a different angle, the angle no one ever thinks of, I had trouble believing they were the same entity.
To give an example, I recall doing an on-site test for a client who had just passed an external test with flying colors. Originally they had only wanted the external test, which was conducted against a handful of IPs. I managed to convince them that in their case, the internal test would provide additional value. I arrived at the office about an hour and a half early, I sat out in the parking lot waiting to go in. I fired up my laptop and noticed a wireless network secured with WEP, the SSID was also the name of the client. You can probably guess what happened next. Four minutes later I had access to the network, and was able to compromise a domain controller via a flaw in some installed backup software. All of this without leaving the car. Eventually, my point of contact arrived and said, “So are you ready to begin, or do you need me to answer some questions first?” The look on his face when I told him that I’d actually already finished was one that I’ll never forget. Just think, had I only performed the external test, I would have been denied that pleasure. Oh, and of course I would have never picked up on the very unsecure wireless network, which is kind of important too.
This is just one example of the kind of thing an internal test can uncover that wouldn’t have even been considered during an external test. Why would an attacker spend several hours scanning a network range when they could just park outside and connect straight to the network?
One of my favorite on-site activities is pretending I’m someone with employee level access gone rogue. Get on the client’s standard build machine with regular user privileges and see how far you can get on the network. Can you install software? Can you load a virtual machine? Can you get straight to the internet, rather than being routed through a proxy? If you can, there are a million and one attack opportunities at your fingertips.
The majority of clients I’ve performed this type of test for hugely overestimated their internal security. It’s well documented that the greatest threat comes from the inside, either on purpose or by accident. But of course, everyone is too busy concentrating on the outside to worry about what’s happening right in front of them.
Good – Networks should be just as hard to break out of, as they are to break in to.
Fortunately, some clients are required to have this type of testing, especially those in government circles. In addition, several IT security auditing standards require a review of internal networks. The depth of these reviews is sometimes questionable though. Auditors aren’t always technical people, and often the review will be conducted against diagrams and documents of how the system is supposed to work, rather than how it actually works. These are certainly useful exercises, but at the end of the day a certificate with a pretty logo hanging from your office wall won’t save you when bad things happen.
Having a remote workforce can be a wonderful thing. You can save a bunch of money by not having to maintain a giant office and the associated IT infrastructure. The downside of this is that in many organizations, the priority is getting people connected and working, rather than properly enforcing security policy. The fact is that if you allow someone to connect remotely into the heart of your network with a machine that you do not have total control over, your network is about as secure as the internet. You are in effect extending your internal network out past the firewall to the unknown. I’ve seen both sides of the spectrum, from an organization that would only allow people to connect in using routers and machines that they configured and installed, to an organization that provided a link to VPN client and said “get on with it”.
I worked with one such client who was starting to rely on remote workers more and more, and had recognized that this could introduce a security problem. They arranged for me to visit the homes of a handful of employees and see if I could somehow gain access to the network’s internal resources. The first employee I visited used his own desktop PC to connect to the network. He had been issued a company laptop, but preferred the big screen, keyboard and mouse that were afforded to him by his desktop. The machine had no antivirus software installed, no client firewall running and no disk encryption. This was apparently because all of these things slowed it down too much. Oh, but it did have a peer-to-peer file sharing application installed. No prizes for spotting the security risks here.
In the second home I visited, I was pleased to see the employee using her company issued XP laptop. Unfortunately she was using it on her unsecured wireless network. To demonstrate why this was a problem, I joined my testing laptop to the network, fired up a Metasploit session and hit the IP with my old favorite, the MS08-067 NetAPI32.dll exploit module. Sure enough, I got a shell, and was able to pivot my way into the remote corporate network. It was at this point that I discovered the VPN terminated in a subnet with unrestricted access to the internal server subnet. When I pointed out to the client that there really should be some sort of segregation between these two areas, I was told that there was. “We use VLAN’s for segregation”, came the response. I’m sure that everyone reading this will know that segregation using VLAN’s, at least from a security point of view, is about as useful as segregating a lion from a Chihuahua with a piece of rice paper. Ineffective, unreliable and will result in an unhappy ending.
Bad – The VPN appliance is located in the core of the network.
We all know that this particular activity is increasing in popularity amongst our adversaries, so why don’t we do it more often as part of our testing? Well, simply put, a lot of the time this comes down to politics. Social engineering tests are a bit of a touchy subject at some organizations, who fear a legal backlash if they do anything to blatantly demonstrate how their own people are subject to the same flaws as the seven billion other on the planet. I’ve been in scoping meetings when as soon as the subject of social engineering has come up, I’m stared at harshly and told in no uncertain terms, “Oh, no way, that’s not what we want, don’t do that.” But why not do it? Don’t you think a malicious hacker would? You’re having a pen test right? Do you think a malicious hacker would hold off on social engineering because they haven’t gotten your permission to try it? Give me a break.
On the other hand, I’ve worked for clients who have recognized the threat of social engineering as one of the greatest to their security, and relished at the opportunity to have their employees tested. Frequently, these tests result in a greater than 80% success rate. So how are they done?
Well, they usually start off with the tester registering a domain name which is extremely similar to the client’s. Maybe with one character different, or a different TLD (“.net” instead of “.com” for example).
The tester’s next step would be to set up a website that heavily borrows CSS code from the client’s site. All it needs is a basic form with username and password fields, as well as some server side coding to email the contents of the form to the tester upon submission.
With messages like this one in an online meeting product, it’s no wonder social engineering attacks are so successful.
Finally, the tester will send out an email with some half-baked story about a new system being installed, or special offers for the employee “if you click this link and login”. Sit back and wait for the responses to come in. Follow these basic steps and within a few minutes, you’ve got a username, password and employee level access. Now all you have to do is find a way to use that to break out of the network, which won’t be too difficult, because everyone will be looking the other way.
The best penetration testers out there are those who provide the best value to the client. This doesn’t necessarily mean the cheapest or quickest. Instead it’s those who make the most effective use of their relatively short window of time, and any other limitations they face to do the job right. Never forget what that job is, and why you are doing it. Sometimes we have to put our generic testing methodologies aside and deliver a truly bespoke product. After all, there is nothing more bespoke than a targeted hacking attack, which can come from any direction. Even from the inside.