Recommended Development Method
We recommend a policy development approach that is consistent with industry best practice as outlined in current standards such as ISO 17799. In general, this approach uses a risk assessment of one's business and technical environment to establish the tolerance for risk and understand minimum or baseline security requirements. This risk analysis enables the identification of threats and countermeasures, and the subsequent development of specifically tailored policy statements.
The following provides an outline of the tasks used to develop security policies.
All responsible organizations and stakeholders are identified and their roles, obligations and tasks detailed.
It is important to understand how your organization is structured, who will be the responsible owner of the security policy and also who will function as its custodian. Most importantly, it is critical to obtain the appropriate level of consensus to ensure that the security policy properly reflects the issues, concerns, requirements, goals, and objectives for your organization. Representation should be as broad as practical but at a minimum include: data security, legal, human resources, internal audit, operations, and development organizations.
The primary business objectives are outlined.
Knowing the primary objectives of your business is important to scoping the security policy effort. For example, one organization may require extensive audit, monitoring, and backup and recovery processes because of regulatory mandates while this may not be applicable to another. The intent here is to make security policy cost effective. That is, do what is appropriate for your organization, not the security consultant selling you the security policies.
A list of security principles representing management's security goals is outlined.
Accompanying this article is a list of security principles. These should be reviewed and incorporated into your security policy development effort as necessary. The purpose of the security principles is to allow your organization to state in a plain and simple fashion, without technical details or jargon, what core values are most important to your organization.
All applicable data and processing resources are identified and classified.
The method that we recommend for security policy development uses a data-centric model. In today's IT environments, data is often one of the most important assets and should be treated accordingly. For that reason, cataloging your data and processing resources enables you to more easily make qualified and informed decisions about their use and value. This then enables you to later apply the most cost effective controls on those assets.
A data flow analysis is performed for the primary data classifications, from generation through deletion.
As noted above, we recommend a data-centric model for policy development. The purpose of a data flow analysis is to allow you to identify all of the trust points that touch your data. For instance, in a transaction processing system, data may flow through browsers, web, data, and other servers or firewalls and be stored in databases, on magnetic tape or paper. By tracing the flow of your data assets through your processing assets, you can later determine the type and placement of logical and physical controls to protect those assets.
The primary threats that can reasonably be expected in one's environment are outlined.
The development of a threat profile enables you to decide what type of threats exist in your particular environment, what the probability is of a threat manifesting itself into an actual problem, and what the ramifications, costs and consequences are of those threats being realized. Remember, threats vary widely between different environments. The threats and consequences of attacks to a financial network processing monetary instruments will be different than the threats and consequences of attacks to an online photo gallery that only displays art.
The primary security services necessary in the environment are identified.
After your data and processing assets are identified and a threat profile created, the next step is to determine what general security services would be appropriate in your environment. These security services are high-level and can include for example: accountability, authorization, availability, identification, authentication, confidentiality, integrity, and non-repudiation. Knowing what security services your environment requires will drive the selection of the types of security policies you will need as well as the specific content or components of those policies.
A generic policy template is constructed.
The structure of a security policy can take many forms. This article offers recommendations for both the components and characteristics of security policies. This step is used to articulate the specific topics that you consider necessary for each security policy. Refer to the Sample Data Security Policy and Guidelines template listed in the References section for some examples of recommended topics.
A list of security policies is defined.
The last step before actually drafting the security policies themselves is to identify all of the security policy focus areas that must be addressed. The creating of this list is based upon the results of the above steps. A sample list of recommended focus areas is included in the Sample Data Security Policy and Guidelines template listed in the References section.
A discussion on the development of security policies requires making certain assumptions. This article presents many data security topics that are presented as examples and should be treated accordingly. Not all subjects are applicable in all circumstances. They are provided to ensure you receive a sufficient body of best practices to enable the development of a comprehensive set of security policies.
Some fundamental security concepts that should be considered when undertaking policy development are described in this section. These concepts are provided as background material to enable you to properly scope your policy development effort.
Underlying all security policy, procedures, and architecture is the expectation that the policy, procedures, and architecture will preserve as confidential that which should remain confidential. In other words, you want to trust the system.
This notion of trust is the ground on which all security stands. In order to create this trust, you must understand and agree to a security policy, and must have confidence that this policy will fulfill their expectations.
In the past, the most common security policy has been no security policy: security was left up to individual implementors, and often overlooked. Following the first damaging security breach, the most common second policy is one so restrictive that it is ignored.
In order to construct a security policy that will neither be overlooked, nor ignored, it is necessary to make certain the security policy reflects realistic business goals and business values. This is done most easily by taking a risk-driven approach.
In finance, risk (R) is defined as:
R = H \ Pe
In this equation, H is the hazard, or the cost of the undesired event, and Pe is the probability of the undesired event. Thus if the cost of an undesired event is large, but the probability of that event is very small, the actual risk is still small.
Consider, for example, the hardened auxiliary power supply for a banking data center. The power supply itself costs $10 million; thus the hazard involved in the loss of the power supply is $10 million (plus whatever opportunity costs might be involved, which for simplicity we ignore here). We estimate the probability of a complete and catastrophic failure of the power supplysay by destruction of the power supply by terroristsas one in a billion, or 1 ∴ 10-9. Thus the risk is $10 million ∴ (1 ∴ 10-9), or $0.01one cent.
On the other hand, consider the personal checking accounts at the same bank: if the bank uses a 4 digit PIN number as the sole means of authentication, then the probability of guessing the PIN correctly in one try (and thereby compromising the account) is 1 ∴ 10-4. If the average checking account balance is $3000, then the risk of the checking account being compromised is $3000 ∴ (1 ∴ 10-4), or $0.30. Surprisingly, the risk to the bank of one checking account being compromised is thirty times the risk of the $10 million dollar power supply being lost.
An easy way to understand this is by considering the cost to the bank of insurance to cover each hazard. An insurance company, in essence, bets its customers that the undesired event will happen rarely enough to allow the insurance company to make a profit on its premium. Since the risk of a checking account being compromised is about 30 cents, this premium has to be slightly more than 30 cents per account.
In general, of course, we don't quantitatively analyze the risk being managed by every specific element of the security policy. An understanding of the nature of risk and how it is quantified, however, can help determine whether any specific element of a security policy is appropriate and suited to the business goals.
It is essential to realize that no one component of a system provides security. Security policies are only effective in the context of an integrated and comprehensive data security architecture; that is, all access control, firewall, cryptographic, key management, and similar mechanisms must work in a coordinated, cohesive and coherent fashion. Thus, in order to provide security, it is not sufficient to simply implement some security features in an IT system. Real security requires us to holistically integrate risk management controls including people, processes, and technology together; security policies must reflect this holistic approach.
Data Centric Approach
The development of security policies is based upon a data centric model. We consider the data of an IT system to be the primary asset at risk. This means that the type of data, its ownership and classification are considered, as well as the flow of data through your organization, systems and networks. Following the data through the system is an effective way to develop efficient and meaningful security policies.