7.4 Security Operations Foundations
7.4 Security Operations Foundations
Security Operations is not only about technology.
Many of the most important operational controls determine who may perform an action, how much authority they receive, whether one person can complete a sensitive process alone and how operational responsibilities are governed.
These controls reduce the risk of mistakes, misuse, fraud, privilege abuse and excessive dependency on individual people or service providers.
Limit
Give people and systems only the access they actually require.
MINIMUM NECESSARYSeparate
Prevent one person from controlling incompatible parts of a sensitive process.
NO SINGLE POINT OF ABUSEGovern
Define responsibilities, privileged access and service expectations.
KNOW WHO OWNS WHATApply Foundational Security Operations Concepts
Limit access to the information and capabilities required to perform authorised duties.
Divide sensitive functions so inappropriate activity cannot be completed by one person acting alone.
Protect and govern accounts capable of performing powerful, security-relevant operations.
Rotate responsibilities among suitably authorised personnel to improve resilience and reduce excessive individual dependency.
Define service responsibilities, expected performance and how service failures or issues are handled.
7.4 Official Topics
The Big Idea
Operational security is stronger when authority is limited, powerful activities are controlled and important responsibilities are not concentrated in one person.
Operations Foundation
Responsibility ยท Authority ยท Accountability
The duty or task a person or function is expected to perform.
WHAT MUST I DO?
The permission or power required to perform that responsibility.
WHAT AM I ALLOWED TO DO?
The ability to associate actions and decisions with the responsible person or function.
WHO ANSWERS FOR IT?
A database administrator may be:
responsible for maintaining production databases,
have authority to execute approved administrative actions,
and be accountable through named accounts and audit logs.
Governance Memory
Need-to-Know & Least Privilege
These concepts are closely related but answer different questions.
Access only the information required to perform an authorised task.
WHICH INFORMATION?
Receive only the minimum capabilities, permissions and resources required to perform an authorised function.
WHICH POWERS?
Need-to-Know vs Least Privilege
๐ฆ Banking Example Same employee - two different restrictions
Fraud investigator.
Can access customer transactions relevant to assigned fraud cases.
Cannot browse:
unrelated celebrity, colleague or family-member accounts.
Can: view and investigate transactions.
Cannot: modify balances or create payments.
Minimum Necessary Access
Least privilege applies to more than ordinary human users.
Receive only permissions required by their duties.
Receive privileged capabilities only for systems they administer.
Application identities receive only required API or database permissions.
Services receive only the system access required for their function.
Workload identities receive narrowly scoped roles.
Devices receive only the network and service access their role requires.
๐ฅ Least Privilege Limits Blast Radius Compromise of one identity should not become compromise of everything
Can access: one application.
Can administer: every server, database, firewall and cloud account.
If both credentials are stolen:
Account B creates dramatically greater potential impact.
Default Deny
A least-privilege model normally starts from the assumption that access is not granted unless there is an authorised requirement.
๐ Privilege Creep Access accumulates while people move through the organisation
Permanent Privilege vs Just-in-Time Privilege
Elevated permission remains assigned continuously.
ALWAYS AVAILABLE
Elevated permission is granted when required and removed or expires afterwards.
AVAILABLE WHEN NEEDED
Normal account: standard access.
Production maintenance: approved privilege for 60 minutes.
After maintenance: privilege expires.
Separation of Duties - SoD
Separation of Duties divides incompatible responsibilities among different people or roles.
The objective is to make it more difficult for one person to misuse a sensitive process without detection or cooperation from someone else.
Separation of Duties
Creating a ยฃ2 Million Payment
Employee A can:
๐ Static vs Dynamic Separation of Duties Prevent conflicting roles permanently or enforce separation during a transaction
Conflicting roles cannot be assigned to the same person.
A user cannot simultaneously hold:
Payment Creator + Payment Approver.
A person may hold several roles but cannot perform conflicting actions within the same transaction or context.
User may create some payments and approve others, but:
cannot approve their own payment.
SoD Types
Separation of Duties vs Dual Control
Divides incompatible responsibilities across different people or functions.
DIVIDE THE PROCESS
Requires two authorised people to participate in or approve a sensitive operation.
TWO PEOPLE REQUIRED
Two authorised custodians must participate before:
a highly sensitive cryptographic operation is performed.
๐งฉ Split Knowledge A related concept commonly used for highly sensitive secrets
Split knowledge divides knowledge of a sensitive value so that no single individual possesses the complete secret.
Secret: divided between multiple authorised custodians.
Three Related Ideas
๐ค Separation of Duties Is Not Perfect Collusion can defeat controls designed around independent people
Employee A creates payment.
Employee B approves payment.
If A and B deliberately cooperate to commit fraud:
Separation of Duties alone may not prevent the activity.
Shared Accounts Create Problems
When several people use the same administrator identity, determining who performed a particular action becomes harder.
User: Administrator
Action: Deleted customer database.
Five people know the password.
The organisation can identify the account but may not be able to identify:
which person actually performed the action.
Privileged Account Management
Privileged accounts can perform security-relevant operations ordinary users cannot perform.
Examples include:
Privileged Access Lifecycle
Privileged Account Lifecycle
๐ค Separate Standard & Administrative Accounts Do not browse the web and administer production with the same identity if avoidable
alice.smith
alice.smith-admin
Protect the Keys to the Kingdom
Prefer attributable administrator identities over unnecessary shared accounts.
Require stronger authentication for high-impact access.
Protect privileged credentials in controlled systems.
Change managed privileged secrets according to policy or usage.
Grant privilege when needed instead of maintaining standing access.
High-risk access may require explicit authorisation.
Record or observe privileged activity where appropriate.
Permit administration only from controlled workstations or networks.
Limit which administrative actions can be performed.
Regularly confirm that elevated access remains necessary.
Record important privileged operations.
Remove temporary elevation when its purpose ends.
๐ Privileged Access Management - PAM Centralised systems can help control powerful accounts and credentials
A PAM capability may help:
Break-Glass Accounts
Organisations may maintain emergency privileged access for situations in which normal authentication or administration mechanisms are unavailable.
Primary identity service fails.
Administrators cannot use normal privileged authentication.
Emergency account provides: controlled recovery access.
Protect Emergency Accounts
๐ค Privileged Service Accounts Not every privileged identity belongs to a human
Service account connects to production database.
Permissions:
database administrator.
Questions to ask:
The Domain Administrator
A systems engineer has permanent domain-administrator privilege and uses the same account for:
A phishing attack steals the engineer's credentials.
Privileged Activity Should Be Visible
Powerful actions deserve stronger monitoring because their consequences can be substantial.
Domain Administrator account added to: another privileged group at 03:12.
Job Rotation
Job rotation periodically moves personnel through different responsibilities or functions.
From a security-operations perspective, it can help reduce excessive dependency on one individual and provide additional oversight of operational processes.
More than one person understands critical processes.
Operations can continue if one employee is unavailable.
A replacement employee may identify unusual practices or concealed activity.
Operational knowledge is less likely to remain with only one person.
Critical processes are less dependent on a single employee.
A different person may identify weaknesses overlooked by the previous operator.
Job Rotation
๐ Job Rotation Must Include Access Rotation Changing jobs without changing permissions can create privilege creep
The Administrator Nobody Can Replace
One administrator has managed a critical financial reconciliation process for:
nine years.
Nobody else understands the process.
The employee:
When another employee finally takes over, they discover:
years of manipulated transactions.
๐๏ธ Related Concept: Mandatory Vacation Often studied alongside job rotation
Mandatory vacation requires personnel in sensitive roles to be absent from their duties for a meaningful period while another authorised person performs the work.
Fraud that requires:
daily manipulation to remain concealed
may become visible while the original employee is absent.
Rotation vs Vacation
Key-Person Dependency
An organisation creates operational risk when a critical process depends entirely on one person's knowledge or availability.
Only one employee knows:
documentation, cross-training, job rotation, controlled credential management and tested procedures.
Service-Level Agreements - SLA
An SLA defines expected service performance and responsibilities between a service provider and its customer.
It converts vague expectations such as:
"The service should be reliable"
into measurable commitments such as:
"The service will provide 99.99% monthly availability."
SLA
What Can an SLA Define?
Which service is being provided?
How much uptime is expected?
Response time, throughput or service quality expectations.
When is support available?
How quickly must the provider acknowledge or respond to an issue?
What restoration or resolution expectations apply?
What belongs to the provider and what belongs to the customer?
What happens if normal support cannot resolve the issue?
Which service metrics and reports must be supplied?
Which security-related service expectations are measurable?
How are maintenance windows handled?
What happens if agreed service levels are not met?
๐ Make Service Levels Measurable "Fast" and "reliable" are difficult to enforce
"Critical incidents will be handled quickly."
"Priority 1 incidents will be acknowledged within 15 minutes, 24 hours a day."
Availability
Availability commitments are often expressed as a percentage over a defined measurement period.
| Availability | Approximate Maximum Downtime per Year |
|---|---|
| 99% | 87 hours 36 minutes |
| 99.9% | 8 hours 46 minutes |
| 99.99% | 53 minutes |
| 99.999% | About 5 minutes |
The SLA should make clear whether scheduled maintenance, customer outages or other exclusions affect availability calculations.
Security-Related Service Levels
Security services can also require measurable operational expectations.
How quickly must the customer be informed?
How quickly must certain security weaknesses be addressed?
What security monitoring service is provided and when?
How quickly will security requests be acknowledged?
Which logs or reports can the customer receive?
Which restoration expectations apply after disruption?
โ๏ธ SLA & Shared Responsibility A provider cannot meet responsibilities it never agreed to own
Responsible for: availability of underlying cloud infrastructure.
Responsible for: correct IAM permissions inside its cloud account.
SLA vs RTO vs RPO
Service commitment between provider and customer.
WHAT LEVEL OF SERVICE IS AGREED?
Recovery Time Objective.
Target time for restoring a disrupted service or capability.
HOW QUICKLY MUST WE RECOVER?
Recovery Point Objective.
Target point in time to which data should be recoverable.
HOW MUCH DATA LOSS CAN WE TOLERATE?
โฑ๏ธ Response Time โ Resolution Time The provider answering the phone is not the same as restoring the service
How quickly the provider acknowledges or begins handling the issue.
How quickly the service or issue is expected to be restored or resolved.
P1 response: 15 minutes.
Target restoration: 4 hours.
SLA Breach
An SLA breach occurs when the agreed service level is not achieved.
Service availability: 99.99%.
Availability: 99.80%.
A service-level failure may be operational, contractual or security related depending on the circumstances.
๐ An SLA Must Be Monitored An agreement without measurement provides limited operational assurance
The Critical Security Provider
A company outsources 24/7 SOC monitoring to a managed security provider.
Contract simply says:
"Provider will monitor security events."
It does not specify:
A Privileged Production Change
A critical firewall rule needs to be changed in production.
Do Not Mix These Concepts
| Concept | Main Question |
|---|---|
| Need-to-Know | Which information do you need? |
| Least Privilege | Which minimum capabilities do you need? |
| Separation of Duties | Which responsibilities should different people perform? |
| Dual Control | Which operation requires two people? |
| Split Knowledge | How can knowledge of a secret be divided? |
| PAM | How do we govern powerful access? |
| Job Rotation | Can more than one person perform the role? |
| SLA | What level of service has been agreed? |
๐ CISSP Scenarios Recognise the foundational Security Operations concept being tested
An employee can see only customer records related to cases assigned to them.
Which principle?
Need-to-know.
A help-desk analyst can reset passwords but cannot create domain administrators.
Which principle?
Least privilege.
A user requires access to one confidential project but can currently view every project.
Primary issue?
Need-to-know violation.
A reporting application needs SELECT permission but is granted full database administrator rights.
Primary issue?
Violation of least privilege.
An employee changes jobs three times and keeps every permission from the previous positions.
Which problem?
Privilege creep.
Elevated access is granted for two hours and then automatically removed.
Which approach?
Just-in-Time privileged access.
One employee can create, approve and release payments.
Which foundational control is missing?
Separation of Duties.
A payment creator is prohibited from ever holding the payment approver role.
Which form of SoD?
Static Separation of Duties.
A manager may create some payments and approve other payments but cannot approve their own.
Which form?
Dynamic Separation of Duties.
Two authorised administrators must participate before a sensitive security operation completes.
Which control?
Dual control.
No one person knows an entire sensitive secret.
Which concept?
Split knowledge.
Two employees with separated responsibilities deliberately cooperate to commit fraud.
Which risk?
Collusion.
Five administrators use one shared root account.
Primary security problem?
Reduced individual accountability.
Each administrator uses a uniquely attributable privileged account.
Which benefit?
Improved accountability and auditing.
A system administrator uses their domain-administrator account for normal web browsing.
Better practice?
Separate ordinary and privileged accounts.
Security wants privileged passwords stored centrally and rotated automatically.
Which capability?
Privileged Access Management - PAM.
An administrator receives production privilege only after approval and only for a maintenance window.
Which concepts?
Least privilege and Just-in-Time access.
A privileged administrative session is recorded for later review.
Why?
Accountability, monitoring and investigation.
The primary identity system fails and administrators cannot authenticate normally.
Which type of account may be required?
Emergency / break-glass account.
A break-glass account is used.
What should normally happen?
Generate visibility, logging and appropriate post-use review.
A service account has domain-administrator privilege even though it only needs access to one database.
Primary issue?
Excessive privilege.
An application account has no known owner.
Which concern?
Privileged identity governance and accountability.
A critical process can be performed by only one employee.
Which operational risk?
Key-person dependency.
Employees periodically exchange operational responsibilities.
Which 7.4 control?
Job rotation.
An employee rotates to a new role but retains privileges from the previous role.
What should have occurred?
Access review and removal of unnecessary permissions.
An employee's hidden fraudulent process becomes visible when another worker temporarily assumes their duties.
Which security benefit?
Job rotation / independent operational exposure.
A sensitive employee is required to take extended leave while another person performs their function.
Which related control?
Mandatory vacation.
A cloud service provider promises 99.99% monthly availability.
Which mechanism defines this commitment?
Service-Level Agreement.
An agreement states that Priority 1 incidents will be acknowledged within 15 minutes.
Which SLA measure?
Response time.
A supplier acknowledges an outage within five minutes but takes ten hours to restore the service.
Which distinction matters?
Response time vs restoration / resolution time.
An SLA says the system must be "highly available" but does not define how availability is calculated.
Primary weakness?
The service level is insufficiently measurable.
A provider promises 99.99% uptime but excludes scheduled maintenance from the calculation.
What should the customer understand?
The SLA's measurement conditions and exclusions.
Service availability falls below the contractual target.
What occurred?
An SLA breach.
Does an SLA breach automatically mean a cyberattack occurred?
Answer?
No.
The cloud provider guarantees infrastructure availability but the customer incorrectly configures its own IAM roles.
Is the provider automatically responsible?
No. Review the shared responsibilities.
A supplier agreement says nothing about how quickly the customer must be notified of security incidents.
Primary concern?
Undefined security service expectation.
Management asks how much data loss the organisation can tolerate following recovery.
Which objective?
RPO.
Management asks how quickly a service must be restored following a disruption.
Which objective?
RTO.
Management asks what performance a service provider has contractually committed to deliver.
Which concept?
SLA.
A provider claims to meet its SLA, but neither side collects the metrics needed to verify performance.
Primary weakness?
Insufficient SLA monitoring and measurement.
One security administrator can create an account, assign privileged roles and approve the access.
Which concern?
Inadequate Separation of Duties.
Privileged access requires a business justification, approval, MFA and automatic expiry.
Which principle is strongly represented?
Controlled least-privilege administration.
A senior executive insists they should have unrestricted access to every system because of their position.
Which principle should govern?
Least privilege and business need.
A DBA has permission to administer databases but cannot access the payroll application because their role does not require the data.
Which concepts?
Least privilege and need-to-know.
An administrator receives approval for an emergency change and then performs it using a named privileged account.
Which property does the named identity strengthen?
Accountability.
Security separates the employee who develops a production script from the employee who approves its deployment.
Which control?
Separation of Duties.
A company has job rotation but keeps operational procedures entirely undocumented.
What remains a concern?
Knowledge transfer and operational resilience.
The organisation removes permanent administrator rights and allows elevation only through an approved workflow.
Primary security improvement?
Reduced standing privilege.
A supplier provides excellent availability but repeatedly fails required security-notification times.
Can an SLA still have been breached?
Yes, if those notification times are agreed service levels.
Management asks what unifies the concepts in 7.4.
Best answer?
Control operational authority, distribute sensitive responsibilities and define accountable service expectations.
Recognise the Clue Words
Only Relevant Data
Information restriction.
Need-to-KnowMinimum Permissions
Capability restriction.
Least PrivilegeOld Permissions Accumulate
Excess access.
Privilege CreepNo Access Unless Approved
Restrictive starting point.
Default DenyTemporary Elevation
Time-limited privilege.
Just-in-TimeCreate โ Approve
Divide functions.
Separation of DutiesCannot Hold Both Roles
Role-level conflict.
Static SoDCannot Approve Own Transaction
Transaction-level conflict.
Dynamic SoDTwo People Required
Shared execution.
Dual ControlSecret Divided
No single holder.
Split KnowledgePeople Cooperate to Defeat SoD
Remaining risk.
CollusionPowerful Administrator Account
Elevated access.
Privileged AccountVault + Rotate Admin Credentials
Control privilege.
PAMDaily User + Admin Identity
Separate them.
Separate Admin AccountEmergency Administrator
Exceptional access.
Break-GlassShared Root Account
Hard to attribute.
Accountability ProblemMove People Between Roles
Cross-training.
Job RotationEmployee Must Be Absent
Fraud exposure.
Mandatory VacationOnly One Person Knows
Resilience issue.
Key-Person Risk99.99% Availability
Service commitment.
SLAProvider Responds Within 15 Minutes
Service metric.
Response TimeRestore Within 4 Hours
Service restoration.
Resolution / Restoration TargetHow Quickly Recover?
Continuity objective.
RTOHow Much Data Loss?
Recovery point.
RPOSupplier Misses Target
Commitment failure.
SLA BreachWho Performs the Work?
Duty.
ResponsibilityWho May Act?
Permission.
AuthorityWho Answers for Action?
Traceability.
Accountabilityโ ๏ธ Common CISSP Mistakes The concepts are related, but they are not interchangeable
Need-to-know limits information.
Least privilege limits capability.
People still need enough access to perform authorised work.
Access should reflect business need and responsibility.
Administrative privilege should still be appropriately scoped.
Non-human identities can hold extremely powerful permissions.
Role changes should trigger access review.
Some elevated access can be granted temporarily.
SoD divides incompatible responsibilities.
Dual control requires multiple people to perform an operation.
Dual control divides action.
Split knowledge divides information.
Collusion can defeat controls based on independent actors.
Knowing which account acted may not reveal which human acted.
Privileged access governance can include approvals, authentication, elevation, monitoring, recording, review and revocation.
Emergency access should be exceptional and controlled.
Emergency use often deserves increased visibility.
Access should change with responsibility.
Personnel must be appropriately trained and authorised for the new responsibilities.
They are related but distinct controls.
Documentation helps, but someone else must also be capable of performing critical work.
An SLA is an agreement.
RTO is a recovery-time objective.
RPO defines tolerated recovery-point loss, not overall service performance.
Responding to an issue does not mean the issue has been resolved.
Availability percentages still permit some outage time.
Responsibilities, measurement, exclusions, reporting and escalation also matter.
Performance should be measured.
Service performance can fail for many reasons.
Understand customer and provider responsibilities.
The customer still needs to understand risk and service performance.
Quick Reference
| If you see... | Think... |
|---|---|
| Only required information | Need-to-Know |
| Only required permissions | Least Privilege |
| Access accumulated over time | Privilege Creep |
| Deny unless specifically required | Default Deny |
| Temporary elevated access | Just-in-Time |
| Creator cannot approve | Separation of Duties |
| Cannot hold conflicting roles | Static SoD |
| Cannot approve own transaction | Dynamic SoD |
| Two people must act | Dual Control |
| Secret divided among people | Split Knowledge |
| Separated people cooperate maliciously | Collusion |
| Elevated administrator capability | Privileged Account |
| Control privileged credentials and sessions | PAM |
| Emergency administrator | Break-Glass Account |
| Everyone uses same admin login | Accountability Problem |
| Move people through responsibilities | Job Rotation |
| Employee required to leave role temporarily | Mandatory Vacation |
| Only one person can operate service | Key-Person Risk |
| Expected provider performance | SLA |
| Acknowledge incident quickly | Response Time |
| Restore service | Resolution / Restoration Time |
| Service uptime percentage | Availability SLA |
| Recover within X hours | RTO |
| Recover data to X point | RPO |
| Provider misses agreed target | SLA Breach |
| What must they do? | Responsibility |
| What may they do? | Authority |
| Who answers for it? | Accountability |
Access Memory Aid
Separation Memory Aid
Privileged Access Memory Aid
Job Rotation Memory Aid
SLA Memory Aid
7.4 Master Memory Aid
Limit โ Separate โ Control โ Rotate โ Agree
The Security Operations Manager's Questions
Key Takeaways
CISSP 7.4 focuses on foundational Security Operations concepts.
The current CISSP outline explicitly includes need-to-know and least privilege, Separation of Duties and responsibilities, privileged account management, job rotation and Service-Level Agreements.
The overall purpose is to control operational authority and reduce the risk created when excessive access, incompatible responsibilities or critical knowledge become concentrated in one place.
Responsibility describes the duty assigned to a person or function.
Authority describes the permission required to perform that duty.
Accountability allows actions and decisions to be associated with the responsible person or function.
Responsibility = duty. Authority = permission. Accountability = answerability.
Need-to-know and least privilege are related but different.
Need-to-know restricts access to information necessary for authorised work.
Least privilege restricts the capabilities and authorisations granted to an identity.
Need-to-know = minimum information. Least privilege = minimum capability.
Least privilege applies to people, administrators, service accounts, applications, workloads, processes and devices.
Excessive privilege increases the potential impact of credential theft, misuse and software compromise.
Privilege determines blast radius.
Default-deny approaches start from no access and grant permissions when a legitimate authorised requirement exists.
Privilege creep occurs when users accumulate permissions while changing roles without old permissions being removed.
Access should therefore be reviewed when responsibilities change.
Just-in-Time privilege can reduce standing administrative access by granting elevated rights only when required and removing them afterwards.
Least privilege can therefore restrict:
what an identity can do, where it can do it and how long the privilege exists.
Separation of Duties divides incompatible activities among different people or functions.
A person who creates a sensitive transaction may therefore be prevented from approving that same transaction.
Static SoD prevents incompatible roles from being assigned to the same identity.
Dynamic SoD can allow several roles while preventing conflicting actions within the same transaction or context.
Static = cannot hold both roles. Dynamic = cannot perform both conflicting actions here.
Dual control and Separation of Duties are related but distinct.
Separation of Duties divides responsibilities.
Dual control requires multiple authorised people to participate in a sensitive operation.
Split knowledge divides knowledge of sensitive information so one person does not possess the entire secret.
SoD = divide duties. Dual control = divide action. Split knowledge = divide information.
Separation of Duties does not eliminate all risk because two authorised individuals can collude.
Logging, monitoring, reconciliation, rotation and independent review can provide additional layers of control.
Shared privileged accounts can weaken accountability because the system may identify which account performed an action without identifying which person used the account.
Unique attributable identities strengthen auditing and accountability.
Privileged accounts can perform security-relevant operations ordinary accounts cannot perform.
Privileged identities may include system administrators, database administrators, cloud administrators, network administrators, service accounts and emergency accounts.
Privileged account = greater authority + greater consequence.
Privileged Account Management should address the lifecycle of privileged access rather than only the storage of passwords.
Controls may include account discovery, approvals, MFA, credential vaulting, rotation, Just-in-Time privilege, session monitoring, access reviews and revocation.
Ordinary and administrative identities should be separated where practical so high privilege is not unnecessarily exposed during routine work such as email and web browsing.
Privileged access should be strongly authenticated and appropriately monitored.
Emergency or break-glass accounts can provide critical access when normal administrative methods fail.
Such accounts should receive strong protection because they can bypass or operate independently of normal access mechanisms.
Emergency access should be available when needed and conspicuous when used.
Service accounts and workload identities can also be privileged.
Non-human identities therefore require ownership, appropriate privilege, credential protection and monitoring.
Job rotation moves appropriately trained personnel through different responsibilities.
It can support cross-training, resilience, knowledge sharing and independent examination of operational practices.
Job rotation can also make irregular behaviour harder to conceal when another person begins performing the process.
Job rotation should be accompanied by access changes.
Rotate the responsibility and reassess the privilege.
Mandatory vacation is a related concept in which an individual in a sensitive function must step away while another authorised person performs the work.
Job rotation and cross-training also reduce key-person dependency.
A critical process that only one employee understands creates both security and resilience risk.
Documentation, cross-training and controlled credential management help distribute operational knowledge.
A Service-Level Agreement defines measurable service expectations and responsibilities between a provider and customer.
SLA terms can include availability, performance, support, response times, restoration expectations, escalation, reporting and security-related service commitments.
SLA = what service, at what level, by whom and measured how.
Service commitments should be measurable.
"Respond quickly" is harder to verify than "acknowledge Priority 1 incidents within 15 minutes."
Availability percentages should be interpreted together with the measurement period and applicable exclusions.
99.99% availability does not mean zero downtime.
Security services can also have measurable service levels, including incident notification and security response commitments.
Outsourced service does not eliminate the customer's need to understand security responsibilities.
Provider and customer obligations should therefore be clearly defined, particularly in cloud and managed-service environments.
Read the responsibility model, not only the uptime percentage.
Response time and resolution time are different.
Response time measures how quickly the provider acknowledges or begins handling an issue.
Resolution or restoration time concerns how quickly the issue or service is restored or resolved.
An SLA is also different from an RTO or RPO.
RTO defines the recovery-time target.
RPO defines the acceptable recovery point for data.
SLA = service commitment. RTO = recovery time. RPO = recovery point.
SLA performance should be monitored rather than merely assumed.
A useful operational process is:
Define โ Measure โ Compare โ Act.
Failure to meet an agreed service target may constitute an SLA breach.
An SLA breach does not automatically mean that a security incident occurred.
The five concepts in 7.4 reinforce each other.
Least privilege limits individual authority.
Separation of Duties distributes sensitive responsibilities.
Privileged Account Management protects elevated authority.
Job rotation distributes knowledge and operational capability.
SLAs define accountable service expectations.
The central CISSP principle is: give people only the access they need, avoid concentrating sensitive authority in one person, closely govern powerful accounts, ensure critical knowledge is not dependent on one individual and make service responsibilities measurable and accountable.
๐ Sources & Further Reading Current and foundational Security Operations references
- ISC2 - CISSP Certification Exam Outline
View the current CISSP Exam Outline - NIST SP 800-53 Rev. 5 - Security and Privacy Controls for Information Systems and Organizations
View NIST SP 800-53 - NIST - Least Privilege
View the NIST least-privilege definition - NIST - Separation of Duty
View the NIST Separation-of-Duty definition - NIST - Privileged Account
View the NIST privileged-account definition - NIST - Privileged Access Management
View the NIST PAM terminology - NIST - Service-Level Agreement
View the NIST SLA definition - NIST SP 800-47 Rev. 1 - Managing the Security of Information Exchanges
View NIST SP 800-47 Rev. 1 - NIST SP 800-35 - Guide to Information Technology Security Services
View NIST security-service guidance
