Penetration Testing methodology
There are logical steps recommended for performing a Penetration Test. The first step is identifying the project's starting status. The most common terminology defining the starting state is Black box testing, White box testing, or a blend between White and Black box testing known as Gray box testing.
Black box assumes the Penetration Tester has no prior knowledge of the target network, company processes, or services it provides. Starting a Black box project requires a lot of reconnaissance and, typically, is a longer engagement based on the concept that real-world attackers can spend long durations of time studying targets before launching attacks.
As a security professional, we find Black box testing presents some problems when scoping a Penetration Test. Depending on the system and your familiarity with the environment, it can be difficult to estimate how long the reconnaissance phase will last. This usually presents a billing problem. Customers, in most cases, are not willing to write a blank cheque for you to spend unlimited time and resources on the reconnaissance phase; however, if you do not spend the time needed then your Penetration Test is over before it began. It is also unrealistic because a motivated attacker will not necessarily have the same scoping and billing restrictions as a professional Penetration Tester. That is why we recommend Gray box over Black box testing.
White box is when a Penetration Tester has intimate knowledge about the system. The goals of the Penetration Test are clearly defined and the outcome of the report from the test is usually expected. The tester has been provided with details on the target such as network information, type of systems, company processes, and services. White box testing typically is focused on a particular business objective such as meeting a compliance need, rather than generic assessment, and could be a shorter engagement depending on how the target space is limited. White box assignments could reduce information gathering efforts, such as reconnaissance services, equaling less cost for Penetration Testing services.
Note
An internal security group usually performs white box testing.
Gray box testing falls in between Black and White box testing. It is when the client or system owner agrees that some unknown information will eventually be discovered during a Reconnaissance phase, but allows the Penetration Tester to skip this part. The Penetration Tester is provided some basic details of the target; however, internal workings and some other privileged information is still kept from the Penetration Tester.
Real attackers tend to have some information about a target prior to engaging the target. Most attackers (with the exception of script kiddies or individuals downloading tools and running them) do not choose random targets. They are motivated and have usually interacted in some way with their target before attempting an attack. Gray box is an attractive choice approach for many security professionals conducting Penetration Tests because it mimics real-world approaches used by attackers and focuses on vulnerabilities rather than reconnaissance.
The scope of work defines how penetration services will be started and executed. Kicking off a Penetration Testing service engagement should include an information gathering session used to document the target environment and define the boundaries of the assignment to avoid unnecessary reconnaissance services or attacking systems that are out of scope. A well-defined scope of work will save a service provider from scope creep (defined as uncontrolled changes or continuous growth in a project's scope), operate within the expected timeframe and help provide more accurate deliverable upon concluding services.
Real attackers do not have boundaries such as time, funding, ethics, or tools meaning that limiting a Penetration Testing scope may not represent a real-world scenario. In contrast to a limited scope, having an unlimited scope may never evaluate critical vulnerabilities if a Penetration Test is concluded prior to attacking desired systems. For example, a Penetration Tester may capture user credentials to critical systems and conclude with accessing those systems without testing how vulnerable those systems are to network-based attacks. It's also important to include who is aware of the Penetration Test as a part of the scope. Real attackers may strike at anytime and probably when people are least expecting it.
Some fundamentals for developing a scope of work for a Penetration Test are as follows:
Definition of Target System(s): This specifies what systems should be tested. This includes the location on the network, types of systems, and business use of those systems.
Timeframe of Work Performed: When the testing should start and what is the timeframe provided to meet specified goals. Best practice is NOT to limit the time scope to business hours.
How Targets Are Evaluated: What types of testing methods such as scanning or exploitation are and not permitted? What is the risk associated with permitted specific testing methods? What is the impact of targets that become inoperable due to penetration attempts? Examples are; using social networking by pretending to be an employee, denial of service attack on key systems, or executing scripts on vulnerable servers. Some attack methods may pose a higher risk of damaging systems than others.
Tools and software: What tools and software are used during the Penetration Test? This is important and a little controversial. Many security professionals believe if they disclose their tools they will be giving away their secret sauce. We believe this is only the case when security professionals used widely available commercial products and are simply rebranding canned reports from these products. Seasoned security professionals will disclose the tools being used, and in some cases when vulnerabilities are exploited, documentation on the commands used within the tools to exploit a vulnerability. This makes the exploit re-creatable, and allows the client to truly understand how the system was compromised and the difficulty associated with the exploit.
Notified Parties: Who is aware of the Penetration Test? Are they briefed beforehand and able to prepare? Is reaction to penetration efforts part of the scope being tested? If so, it may make sense not to inform the security operations team prior to the Penetration Test. This is very important when looking at web applications that may be hosted by another party such as a cloud service provider that could be impacted from your services.
Initial Access Level: What type of information and access is provided prior to kicking off the Penetration Test? Does the Penetration Tester have access to the server via Internet and/or Intranet? What type of initial account level access is granted? Is this a Black, White, or Gray box assignment for each target?
Definition of Target Space: This defines the specific business functions included in the Penetration Test. For example, conducting a Penetration Test on a specific web application used by sales while not touching a different application hosted from the same server.
Identification of Critical Operation Areas: Define systems that should not be touched to avoid a negative impact from the Penetration Testing services. Is the active authentication server off limits? It's important to make critical assets clear prior to engaging a target.
Definition of the Flag: It is important to define how far a Penetration Test should compromise a system or a process. Should data be removed from the network or should the attacker just obtain a specific level of unauthorized access?
Deliverable: What type of final report is expected? What goals does the client specify to be accomplished upon closing a Penetration Testing service agreement? Make sure the goals are not open-ended to avoid scope creep of expected service. Is any of the data classified or designated for a specific group of people? How should the final report be delivered? It is important to deliver a sample report or periodic updates so that there are no surprises in the final report.
Remediation expectations: Are vulnerabilities expected to be documented with possible remediation action items? Who should be notified if a system is rendered unusable during a Penetration Testing exercise? What happens if sensitive data is discovered? Most Penetration Testing services do NOT include remediation of problems found.
Some service definitions that should be used to define the scope of services are:
Security Audit: Evaluating a system or an application's risk level against a set of standards or baselines. Standards are mandatory rules while baselines are the minimal acceptable level of security. Standards and baselines achieve consistency in security implementations and can be specific to industries, technologies, and processes.
Most requests for security serves for audits are focused on passing an official audit (for example preparing for a corporate or a government audit) or proving the baseline requirements are met for a mandatory set of regulations (for example following the HIPAA and HITECH mandates for protecting healthcare records). It is important to inform potential customers if your audit services include any level of insurance or protection if an audit isn't successful after your services. It's also critical to document the type of remediation included with audit services (that is, whether you would identify a problem, offer a remediation action plan or fix the problem). Auditing for compliance is much more than running a security tool. It relies heavily on the standard types of reporting and following a methodology that is an accepted standard for the audit.
In many cases, security audits give customers a false sense of security depending on what standards or baselines are being audited. Most standards and baselines have a long update process that is unable to keep up with the rapid changes in threats found in today's cyber world. It is HIGHLY recommended to offer security services beyond standards and baselines to raise the level of security to an acceptable level of protection for real-world threats. Services should include following up with customers to assist with remediation along with raising the bar for security beyond any industry standards and baselines.
Vulnerability Assessment: This is the process in which network devices, operating systems and application software are scanned in order to identify the presence of known and unknown vulnerabilities. Vulnerability is a gap, error, or weakness in how a system is designed, used, and protected. When a vulnerability is exploited, it can result in giving unauthorized access, escalation of privileges, denial-of-service to the asset, or other outcomes.
Vulnerability Assessments typically stop once a vulnerability is found, meaning that the Penetration Tester doesn't execute an attack against the vulnerability to verify if it's genuine. A Vulnerability Assessment deliverable provides potential risk associated with all the vulnerabilities found with possible remediation steps. There are many solutions such as Kali Linux that can be used to scan for vulnerabilities based on system/server type, operating system, ports open for communication and other means. Vulnerability Assessments can be White, Gray, or Black box depending on the nature of the assignment.
Vulnerability scans are only useful if they calculate risk. The downside of many security audits is vulnerability scan results that make security audits thicker without providing any real value. Many vulnerability scanners have false positives or identify vulnerabilities that are not really there. They do this because they incorrectly identify the OS or are looking for specific patches to fix vulnerabilities but not looking at rollup patches (patches that contain multiple smaller patches) or software revisions. Assigning risk to vulnerabilities gives a true definition and sense of how vulnerable a system is. In many cases, this means that vulnerability reports by automated tools will need to be checked.
Customers will want to know the risk associated with vulnerability and expected cost to reduce any risk found. To provide the value of cost, it's important to understand how to calculate risk.
Calculating risk
It is important to understand how to calculate risk associated with vulnerabilities found, so that a decision can be made on how to react. Most customers look to the CISSP triangle of CIA when determining the impact of risk. CIA is the confidentiality, integrity, and availability of a particular system or application. When determining the impact of risk, customers must look at each component individually as well as the vulnerability in its entirety to gain a true perspective of the risk and determine the likelihood of impact.
It is up to the customer to decide if the risk associated to vulnerability found justifies or outweighs the cost of controls required to reduce the risk to an acceptable level. A customer may not be able to spend a million dollars on remediating a threat that compromises guest printers; however, they will be very willing to spend twice as much on protecting systems with the company's confidential data.
The Certified Information Systems Security Professional (CISSP) curriculum lists formulas for calculating risk as follow.
A Single Loss Expectancy (SLE) is the cost of a single loss to an Asset Value (AV). Exposure Factor (EF) is the impact the loss of the asset will have to an organization such as loss of revenue due to an Internet-facing server shutting down. Customers should calculate the SLE of an asset when evaluating security investments to help identify the level of funding that should be assigned for controls. If a SLE would cause a million dollars of damage to the company, it would make sense to consider that in the budget.
The Single Loss Expectancy formula:
SLE = AV * EF
The next important formula is identifying how often the SLE could occur. If an SLE worth a million dollars could happen once in a million years, such as a meteor falling out of the sky, it may not be worth investing millions in a protection dome around your headquarters. In contrast, if a fire could cause a million dollars worth of damage and is expected every couple of years, it would be wise to invest in a fire prevention system. The number of times an asset is lost is called the Annual Rate of Occurrence (ARO).
The Annualized Loss Expectancy (ALE) is an expression of annual anticipated loss due to risk. For example, a meteor falling has a very low annualized expectancy (once in a million years), while a fire is a lot more likely and should be calculated in future investments for protecting a building.
Annualized Loss Expectancy formula:
ALE = SLE * ARO
The final and important question to answer is the risk associated with an asset used to figure out the investment for controls. This can determine if and how much the customer should invest into remediating vulnerability found in a asset.
Risk formula:
Risk = Asset Value * Threat * Vulnerability * Impact
It is common for customers not to have values for variables in Risk Management formulas. These formulas serve as guidance systems, to help the customer better understand how they should invest in security. In my previous examples, using the formulas with estimated values for a meteor shower and fire in a building, should help explain with estimated dollar value why a fire prevention system is a better investment than metal dome protecting from falling objects.
Penetration Testing is the method of attacking system vulnerabilities in a similar way to real malicious attackers. Typically, Penetration Testing services are requested when a system or network has exhausted investments in security and clients are seeking to verify if all avenues of security have been covered. Penetration Testing can be Black, White, or Gray box depending on the scope of work agreed upon.
The key difference between a Penetration Test and Vulnerability Assessment is that a Penetration Test will act upon vulnerabilities found and verify if they are real reducing the list of confirmed risk associated with a target. A Vulnerability Assessment of a target could change to a Penetration Test once the asset owner has authorized the service provider to execute attacks against the vulnerabilities identified in a target. Typically, Penetration Testing services have a higher cost associated since the services require more expensive resources, tools, and time to successfully complete assignments. One popular misconception is that a Penetration Testing service enhances IT security since services have a higher cost associated than other security services:
Penetration Testing does not make IT networks more secure, since services evaluate existing security! A customer should not consider a Penetration Test if there is a belief the target is not completely secure.
Penetration Testing can cause a negative impact to systems: It's critical to have authorization in writing from the proper authorities before starting a Penetration Test of an asset owned by another party. Not having proper authorization could be seen as illegal hacking by authorities. Authorization should include who is liable for any damages caused during a penetration exercise as well as who should be contacted to avoid future negative impacts once a system is damaged. Best practice is alerting the customers of all the potential risks associated with each method used to compromise a target prior to executing the attack to level set expectations. This is also one of the reasons we recommend targeted Penetration Testing with a small scope. It is easier to be much more methodical in your approach. As a common best practice, we receive confirmation, which is a worst case scenario, that a system can be restored by a customer using backups or some other disaster recovery method.
Penetration Testing deliverable expectations should be well defined while agreeing on a scope of work. The most common methods by which hackers obtain information about targets is through social engineering via attacking people rather than systems. Examples are interviewing for a position within the organization and walking out a week later with sensitive data offered without resistance. This type of deliverable may not be acceptable if a customer is interested in knowing how vulnerable their web applications are to remote attack. It is also important to have a defined end-goal so that all parties understand when the penetration services are considered concluded. Usually, an agreed-upon deliverable serves this purpose.
A Penetration Testing engagement's success for a service provider is based on profitability of time and services used to deliver the Penetration Testing engagement. A more efficient and accurate process means better results for less services used. The higher the quality of the deliverables, the closer the service can meet customer expectation, resulting in a better reputation and more future business. For these reasons, it's important to develop a methodology for executing Penetration Testing services as well as for how to report what is found.
