The Value of Honeypots
- Jan 10, 2003
Now that we have defined honeypots and how they work, we can attempt to establish their value. As mentioned earlier, unlike mechanisms such as firewalls and intrusion detection systems, a honeypot does not address a specific problem. Instead, it is a tool that contributes to your overall security architecture. The value of honeypots and the problems they help solve depend on how you build, deploy, and use them.
Honeypots have certain advantages and disadvantages that affect their value. In this chapter we will examine those advantages and disadvantages more closely. We will also look at the differences between production and research honeypots and their respective roles.
Advantages Of Honeypots
Honeypots have several advantages unique to the technology. We will review four of them here.
One of the challenges the security community faces is gaining value from data. Organizations collect vast amounts of data every day, including firewall logs, system logs, and Intrusion Detection alerts. The sheer amount of information can be overwhelming, making it extremely difficult to derive any value from the data. Honeypots, on the other hand, collect very little data, but what they do collect is normally of high value. The honeypot concept of no expected production activity dramatically reduces the noise level. Instead of logging gigabytes of data every day, most honeypots collect several megabytes of data per day, if even that much. Any data that is logged is most likely a scan, probe, or attackinformation of high value.
Honeypots can give you the precise information you need in a quick and easy-to-understand format. This makes analysis much easier and reaction time much quicker. For example, the Honeynet Project, a group researching honeypots, collects on average less then 1MB of data per day. Even though this is a very small amount of data, it contains primarily malicious activity. This data can then be used for statistical modeling, trend analysis, detecting attacks, or even researching attackers. This is similar to a microscope effect. Whatever data you capture is placed under a microscope for detailed scrutiny.
For example, in Figure 4-1 we see a scan attempt made against a network of hon-eypots. Since honeypots have no production value, any connection made to a honeypot is most likely a probe or attack. Also, since such little information is collected, it is very easy to collate and identify trends that most organizations would miss. In this figure we see a variety of UDP connections made from several systems in Germany. At first glance, these connections do not look related, since different source IP addresses, source ports, and destination ports are used. However, a closer look reveals that each honeypot was targeted only once by these different systems. Analysis reveals that an attacker is doing a covert network sweep.
Figure 4-1 Covert network sweep by an attacker picked up by a network of honeypots
He is attempting to determine what systems are reachable on the Internet by sending UDP packets to high ports, similar to how traceroute works on Unix. Most systems have no port listening on these high UDP ports, so when a packet is sent, the target systems send an ICMP port unreachable error message. These error messages tell the attacker that the system is up and reachable.
The attacker makes this network sweep difficult to detect because he randomizes the source port and uses multiple source IP addresses. In reality, he is most likely using a single computer for the scan but has aliased multiple IP addresses on the system or is sniffing the network for return packets to the different systems. Organizations that collect large amounts of data would most likely miss this sweep, since multiple-source IP addresses and source ports make it hard to detect. However, because honeypots collect small amounts of, but high-value data, attacks like these are extremely easy to identify. This demonstrates one of the most critical advantages of honeypots.
Another challenge most security mechanisms face is resource limitations, or even resource exhaustion. Resource exhaustion is when a security resource can no longer continue to function because its resources are overwhelmed. For example, a firewall may fail because its connections table is full, it has run out of resources, or it can no longer monitor connections. This forces the firewall to block all connections instead of just blocking unauthorized activity. An Intrusion Detection System may have too much network activity to monitor, perhaps hundreds of megabytes of data per second. When this happens, the IDS sensor's buffers become full, and it begins dropping packets. Its resources have been exhausted, and it can no longer effectively monitor network activity, potentially missing attacks. Another example is centralized log servers. They may not be able to collect all the events from remote systems, potentially dropping and failing to log critical events.
Because they capture and monitor little activity, honeypots typically do not have problems of resource exhaustion. As a point of contrast, most IDS sensors have difficulty monitoring networks that have gigabits speed. The speed and volume of the traffic are simply too great for the sensor to analyze every packet. As a result, traffic is dropped and potential attacks are missed. A honeypot deployed on the same network does not share this problem. The honeypot only captures activities directed at itself, so the system is not overwhelmed by the traffic. Where the IDS sensor may fail because of resource exhaustion, the honeypot is not likely to have a problem. A side benefit of the limited resource requirements of a honeypot is that you do not have to invest a great deal of money in hardware for a honeypot. Honeypots, in contrast to many security mechanisms such as firewalls or IDS sensors, do not require the latest cutting-edge technology, vast amounts of RAM or chip speed, or large disk drives. You can use leftover computers found in your organization or that old laptop your boss no longer wants. This means that not only can a honeypot be deployed on your gigabit network but it can be a relatively cheap computer.
I consider simplicity the biggest single advantage of honeypots. There are no fancy algorithms to develop, no signature databases to maintain, no rulebases to misconfigure. You just take the honeypot, drop it somewhere in your organization, and sit back and wait. While some honeypots, especially research honey-pots, can be more complex, they all operate on the same simple premise: If somebody or someone connects to the honeypot, check it out. As experienced security professionals will tell you, the simpler the concept, the more reliable it is. With complexity come misconfigurations, breakdowns, and failures.
Return On Investment
When firewalls successfully keep attackers out, they become victims of their own success. Management may begin to question the return on their investment, as they perceive there is no longer a threat: "We invested in and deployed a firewall three years ago, and we were never attacked. Why do we need a firewall if we have never been hacked?" The reason they were never hacked is the firewall helped reduce the risk. Investments in other security technologies, such as strong authentication, encryption, and host-based armoring, face the same problem. These are expensive investments, costing organizations time, money, and resources, but they can become victims of their own success.
In contrast, honeypots quickly and repeatedly demonstrate their value. Whenever they are attacked, people know the bad guys are out there. By capturing unauthorized activity, honeypots can be used to justify not only their own value but investments in other security resources as well. When management perceives there are no threats, honeypots can effectively prove that a great deal of risk does exist.
For example, once I was in Southeast Asia conducting a security assessment for a large financial organization. I was asked to do a presentation for the Board of Directors on the state of their security. As always, I had a honeypot running on my laptop. About 30 minutes before the presentation, I connected to their network to make some last-minute changes. Sure enough, while I was connected to their network, my system was probed and attacked. Fortunately, the honeypot captured the entire attempt. In this case the attack was a Back Orifice scan. When the attacker found my system, they thought it was infected and executed a variety of attacks, including attempting to steal my password and reboot the system. I then went with this captured attack and used it to open my presentation to the Board. This attack demonstrated to the Board members that not only did active threats exist but they tried, and succeeded, in penetrating their network. It is one thing to talk about such threats, but demonstrating them, keystroke by keystroke, is far more effective. This proved extremely valuable in getting the Board's attention.