Privileged Access Management

How to control administrator, break-glass, and sensitive operational access with just-in-time privilege, approval, logging, and review.

Privileged access is access that can materially change system security, availability, data integrity, or evidence. It includes administrator roles, key-management rights, production deployment authority, identity administration, audit-log control, database owner rights, and emergency break-glass access.

The point of privileged access management is not to make administrators miserable. The point is to ensure that powerful access is justified, time-bounded when possible, individually attributable, strongly authenticated, and visible afterward.

Privileged access controls

Control What it reduces
Dedicated admin accounts Accidental use of privileged rights during normal work
MFA or phishing-resistant authentication Account takeover risk for powerful identities
Just-in-time elevation Standing privilege and long exposure windows
Approval workflow Unauthorized or unnecessary privileged sessions
Session logging or command recording Loss of accountability during sensitive actions
Privilege separation One actor controlling too many incompatible steps
Break-glass controls Inability to respond when normal identity systems fail
Access review Privilege drift after role changes or incidents

The strongest design usually combines several controls. MFA alone does not justify permanent global administrator rights.

Just-in-time access flow

    flowchart LR
	  A["Admin requests elevation"] --> B["Policy checks role, risk, and reason"]
	  B --> C["Approval or automated eligibility"]
	  C --> D["Temporary privileged session"]
	  D --> E["Action logged and reviewed"]
	  E --> F["Privilege expires"]

This pattern narrows time while preserving accountability. It is especially strong for maintenance windows, incident response, production deployments, and sensitive data access.

Break-glass access

Break-glass accounts or procedures exist for emergencies: identity-provider outage, federation misconfiguration, lost administrator access, or severe operational incident. They should be:

  • rare and documented
  • protected with strong authentication and secure custody
  • excluded from routine daily operations
  • monitored with high-signal alerts
  • tested periodically
  • reviewed immediately after use

A break-glass account that becomes a daily shared admin login is no longer emergency control. It is unmanaged standing privilege.

Root, owner, and global admin accounts

Every platform has a highest-risk authority: cloud root user, tenant global administrator, domain administrator, cluster-admin, database owner, or organization owner. Strong answers typically:

  • minimize routine use
  • protect with strong authentication
  • remove unnecessary access keys or persistent credentials
  • create lower-privilege roles for daily work
  • log every use
  • separate billing, identity, security, and workload administration where possible

When an exam option uses the highest-privilege account for routine work, treat it with suspicion.

Privileged access and evidence

Privileged actions must be attributable. Useful evidence includes:

  • principal identity and session identity
  • authentication method and MFA context
  • approval ticket or reason
  • source device, network, and location where relevant
  • exact command, API action, or configuration change
  • before-and-after state for critical changes
  • session end time and revocation status

Shared administrator accounts, generic service accounts, and copied credentials weaken evidence even when they appear convenient.

Common traps

Trap Corrective rule
“All admins need permanent broad access.” Use eligibility, JIT elevation, scoped roles, and emergency fallback.
“MFA is enough for global admin.” MFA is one control; privilege scope, time, logging, and review still matter.
“Shared admin accounts simplify operations.” They break accountability and deprovisioning.
“Break-glass should be easy for everyone.” It should be available under emergency procedure, not casual access.
“Disable logging during incident response.” Incidents increase the need for evidence.

Sample Exam Question

A security team wants database administrators to perform emergency maintenance without keeping permanent owner rights active all month. The organization also needs evidence of who used elevated access. Which design is strongest?

A. A shared database owner account stored in a password manager
B. Permanent owner rights for every database administrator
C. Just-in-time privileged access with individual identities, approval or eligibility policy, MFA, expiration, and audit logging
D. A public maintenance account with a rotating password

Best answer: C. It narrows standing privilege, preserves individual accountability, and produces usable evidence.

Revised on Friday, September 11, 2026