Home > Articles > Security > General Security and Privacy

Security Through Penetration Testing: Internet Penetration

📄 Contents

  1. 1 Network Enumeration/Discovery
  2. 2 Vulnerability Analysis
  3. 3 Exploitation
  • Print
  • + Share This
The overall methodology for penetration testing can be broken into a three-step process: network enumeration, vulnerability analysis, and exploitation. That means discovering as much as possible about the target, identifying all potential avenues of attack, and attempting to compromise the network by leveraging the results of the vulnerability analysis and following as many avenues identified as time allows. Throughout the discussion of this process, there are references to the tools found most useful for accomplishing these tasks.
This chapter is from the book

This chapter begins our discussion of the general process for performing penetration testing that we have developed during our experience. While the procedures discussed are not set in stone and we never cease to examine and refine our own techniques, we would like to stress that the approach laid out is both an efficient means of compromising a network and an effective means of evaluating the security posture of that network.

That is not to say it is the only means of examining the security posture of a network. Other security professionals have different and valid testing techniques. This process is one that has proven to be effective.

Having a defined, organized methodology provides for an efficient penetration test with a consistent level of detail. Professional consultants hired to perform penetration testing attempt to compromise the target network during a given time period, often a matter of weeks or even days. This is substantially different than hackers who can spend as much time as they want in attempting to gain root access to a network. Therefore, we need a well-defined methodology that allows us to systematically check for known vulnerabilities and pursue potential security holes in the time allotted. In addition, following a single methodology helps ensure a consistent level of reliability in results across multiple engagements.

The overall methodology for penetration testing can be broken into a three-step process.

  1. Network enumeration: Discover as much as possible about the target.

  2. Vulnerability analysis: Identify all potential avenues of attack.

  3. Exploitation: Attempt to compromise the network by leveraging the results of the vulnerability analysis and following as many avenues identified as time allows.

Throughout our discussion of this process, we reference the tools we have found most useful for accomplishing these tasks.

5.1 Network Enumeration/Discovery

Before we can gain unauthorized access to a network, we have to know the topology of the network. Every piece of information we can obtain about the target network adds a piece to the puzzle. We specifically scan the target network to obtain a list of live hosts, as well as to begin mapping the target to get a sense of its architecture and the kind of traffic (for example, TCP, UDP, IPX) that is allowed. The goal of discovery is to start with no information and gather as much data as possible about the target network and systems. We then use this information to identify potential exploits.

The process of discovering this information is called network enumeration and is the first step to an external penetration test. This step is performed largely over the Internet using readily available software and publicly accessible repositories of information. Most of the information we obtain in this step is freely available and legal to obtain. However, many companies monitor who tries to get this information since it may indicate a prelude to an attack.

5.1.1 Whois Query

Even before we begin the network scanning, we must determine the domain names and IP address ranges that belong to the target organization. To simulate the scenario of an external hacker, no prior information about the target organization should be provided to the consultant to best determine the amount of information a hacker could obtain. However, before moving to the second step of the process, all identified domain names and IP addresses should be verified with the target organization to ensure they are owned by the organization and are part of the scope of the exercise.

To determine the IP address ranges associated with the client, we perform an Internet whois query. The command can be run natively on most UNIX environments (check man whois for usage and version-specific syntax). For the Windows environment, Ws PingPro Pack and Sam Spade are two tools that can be used to perform whois queries. (These tools are discussed in Chapter 12.)

Whois queries can also be made over the Web from http://www.arin.net and http://www.networksolutions.com. Figure 5–1 shows the whois query from the Network Solutions site (without the domain servers) for the domain klevinsky.com.

Figure 5–1 Whois query for klevinsky.com

A whois query provides the administrative contact, billing contact, and address of the target network. The administrative and billing contact information can be useful for performing social engineering attacks on the employees of the target network (see Chapter 8).

The whois query provides IP address ranges that are associated with the name you enter. Some ranges may be returned that belong to a separate organization with a similar name. For example, the partial results of a whois query on company reveal registered IP addresses for a collection of firms whose names include the word company but may not be the target organization.

Of the multiple IP ranges that do belong to the client, a portion may belong to different divisions of the client's organization and lie outside the scope of the engagement. The targets for the engagement should be verified when this information is found.

Whois queries return only the first 50 items that match the query. This is implemented by Internic to limit the search time. As the listings of Internet domains grow, the task of searching all listings and returning all possible matches becomes more computationally intensive.

If the target company has more than 50 listings that interest you, you may have to engage in some creative searching. One idea is to break up the names of the company or search for plurals or modified company names. Find the names of subsidiary organizations (press releases on the target company's Web site are a good place to look) and search for those names as well.

5.1.2 Zone Transfer

A whois query also returns the list of domain name servers that provide the target network's host name and IP address mapping. (This information, along with the contact information, is found by clicking on the Net Block name associated with the listing.) To obtain the network IP listing, we want to attempt a zone transfer against each system identified as a DNS server. A zone transfer requests the complete list of matched IP addresses and host names stored within a DNS for a specified domain.

A zone transfer can be performed with the nslookup command that is supported by both the UNIX and Windows platforms. Sam Spade, Ws PingPro Pack, and NetScan Tools on the Windows operating system all provide a graphical user interface (GUI) for performing a zone transfer. In order to perform a zone transfer, we have to use a DNS server that is authoritative for the domain of interest; therefore, we use the domain name servers identified through the whois query. Techniques for performing zone transfers are covered in Chapter 12.

The zone transfer returns a listing of IP addresses and their corresponding host names. A typical listing may look something like this:

ls -d abc.com
[server.abc.com]
 abc.com.           SOA        server.abc.com
admin.abc.com.  (200000068 300 800 359100 4700)
 abc.com.           A          10.10.10.30
 abc.com.           NS         server.abc.com
 abc.com.           MX         10 mail.abc.com
 business           A          10.10.10.11
 application        A          10.10.10.32
 mailsweeper        A          10.10.10.50
 mimesweeper        CNAME      server4.abc.com
 server4            A          10.10.10.40
 abc.com.           SOA        server.abc.com
admin.abc.com.  (200000068 300 800 359100 4700)

Machine host names often indicate the function of the machine. For instance, the corporate firewall machine is often called "firewall" or the name of the firewall running, such as "Gauntlet" or "Firewall1." Similarly, we have seen some equally revealing machine names, such as "mail.companyname.com," "smtp.companyname.com," "ftp.companyname.com," "dns01.companyname.com," "ns01.companyname.com," and "web03.companyname.com." These names not only offer strong evidence of their main function but also indicate the presence of other machines. For example, if there is a web03 machine on a particular network, there stands to reason that a web01 and a web02 may also exist. If there is an ns01 machine, there may also be ns and ns02 machines. In light of this, names of sports teams, famous people, and cartoon characters have been used as good machine names. They are easy to remember, and they do not give away any technical information.

When doing a zone transfer, keep in mind that often the DNS server does not have a complete listing for all the target network's hosts. Several machines may be using DHCP, and the company may use separate domain name servers for separate domains. Also, its DNS may not support zone transfer requests from unauthorized hosts, allowing them only from the backup name servers within the organization. Therefore, you should attempt zone transfers against all the target network's identified domain name servers. One may offer at least a partial listing.

We have also seen companies outsource the domain name function or use their ISP's DNS server. In our experience, performing a zone transfer against a DNS server or any machine belonging to an ISP or a third party is generally not received well by those third parties. In that case, we usually omit this step unless we have the written consent of both the target organization and the third party. In these situations, make sure the terms of the penetration test clearly state whether or not the hosted systems are within the scope of the engagement.

On the other hand, DNS machines that belong to the client organization but are not a part of the IP address range are specifically within scope and are valid targets of a zone transfer as long as there is a reasonable chance that that DNS will offer information regarding the within-scope target domain. This is because an Internet-based penetration relies on using information that lies in the public domain or is publicly accessible.

This usually occurs when the target comprises one or more domains within a large organization. The main DNS server for the organization will likely have a partial listing of the hosts in the target domain even if it lies outside that domain.

Unlike the whois query, a zone transfer is fairly indicative of hacker activity since there really is no need for the general user to have this information. Therefore, someone making this query against a DNS server is probably a potential attacker. For that reason, we suggest exercising good judgment before performing these queries. Zone transfers may indicate to the network staff the beginning of a penetration test against the network.

5.1.3 Ping Sweeps

Our next step is to ping the discovered IP addresses to see if they are "up" or "live." There are a variety of ways to ping a set of IP addresses. The most commonly used is the traditional ICMP ping (with echo requests or echo replies messages), but gaining popularity is a TCP ping (with a full or half TCP handshake). Many sites have taken the security step of restricting ICMP traffic or blocking it at the border firewall and router, limiting their exposure to the traditional ping. However, a TCP ping may still be allowed on the network.

Over time, organizations have become more adept at blocking a ping sweep, and countermeasures are becoming more prevalent. While you can assume with some amount of confidence that a host that sends an ICMP response to an ICMP echo request is active, it is not always true that a host that fails to send such a response is necessarily down. The host may be down, or ICMP traffic to that host may be filtered and the ping request simply did not reach it. False responses can also be sent to ICMP echo requests by perimeter security devices.

Depending on the level of stealth you are seeking in your pinging activity, there are a variety of steps you can take to remain beneath the radar of an intrusion detection system that may be monitoring network traffic. While these steps are discussed in greater detail in the section on Nmap in Chapter 12, it is worth mentioning that randomizing the order of the IP addresses being pinged helps avoid detection, as do varying the time between sending ping packets and dividing the IP addresses into multiple groups (this is most helpful for large numbers of hosts, that is, over 100).

The ping utility exists natively on most operating systems and can be performed from a large collection of tools. One of the most popular is Nmap because of its configuration, its ease of use, and the other features it includes (TCP ping, port scanning, OS identification). For the Windows environments, Pinger and Ws PingPro Pack are both effective tools for performing ping sweeps. (In addition, a Windows-compatible version on Nmap is currently under development.) Pinger strictly pings a set of IP addresses while Ws PingPro Pack provides additional functionality through a suite of tools.

Ping sweeps are generally not considered to be evidence of harmful intent to hack a system. However, they can be irritating or destructive if they become excessive; for example, ping each box on a Class C network every 30 seconds for 8 hours and see how that affects bandwidth.

5.1.4 Traceroute

In order to come up with a rough map of the client architecture, we trace the route to several of the live hosts. This is a tedious process, but it does help identify the routers, firewalls, load-balancing devices, and other border machines in place on the target network. In addition, it helps identify hosts that are on separate segments. Hosts on separate segments may be managed by different individuals and may have trust relationships that can be exploited to compromise the system.

A traceroute marks the path of ICMP packets from the local host (where the command is executed) to the destination host. It is available as a command line tool on both the UNIX (traceroute) and Windows (tracert) operating systems. In addition, the Windows-based tool VisualRoute performs this service as well as mapping the path over a map of the world. (VisualRoute is discussed in Chapter 12.)

We perform traceroutes on several IP addresses within the same Class C address block to see if the ICMP packets follow the same path. We are interested in seeing the hops just prior to the target. These hops may represent routers, firewalls, or other gateways. If several hosts have the same prior hop, it is probably a router or firewall. If there is a common host after which ICMP packets can no longer be seen, that too may be the firewall or filtering router. Also, a common host in front of a bank of Web servers may be a load-balancing device or a Web redirector.

If you notice that packets to some hosts on the network segment follow an alternate path, you may have discovered new gateways into the target network. It is not uncommon for network segments to have multiple connections to the Internet—unbeknownst to network managers. These can be developed on the fly for particular network tests or applications and simply forgotten. Such paths often lead to network compromises.

  • + Share This
  • 🔖 Save To Your Account