Design Access Control System - System Design | WinJob

Problem Statement & Authorization Control Plane

Enterprise access control is the authorization control plane: it decides whether principal P may perform action A on resource R after authentication has already established identity. RBAC/ABAC platform separating authentication from fine-grained authorization across microservices.

Key points

AuthZ answers what a principal may do on which resource instance

PEP at edge enforces; PDP evaluates versioned policy bundles

Deny-first evaluation prevents privilege escalation by default

Deep dive

Interviewers at AWS IAM, Google Zanzibar, and Okta-class platforms expect you to articulate evaluation order, cache strategy, and auditability. For section 1, stress how problem statement & authorization control plane reduces privilege creep: every grant is attributable, every deny is explainable, and policy changes roll forward through immutable bundle versions rather than mutating rows in place.

When sizing Problem Statement & Authorization Control Plane, separate admin write QPS (low) from IsAuthorized read QPS (massive). A single listing page may issue fifty batch checks; without PEP caching the PDP becomes a synchronous bottleneck. Document fail-closed behavior for privileged routes when PDP latency exceeds SLO or health checks fail.

Implementation sketch

javaOne Dark Pro

1 public final class AuthzCtx1 {
2   private final String tenantId;
3   private final long bundleVersion;
4   private final String resourceUrn;
5   public boolean isStale(long live) { return bundleVersion < live; }
6   public String cacheKey(String subject, String action) {
7     return tenantId + "|" + bundleVersion + "|" + subject + "|" + action + "|" + resourceUrn.hashCode();
8   }
9 }

pythonOne Dark Pro

1 from dataclasses import dataclass
2  
3 @dataclass(frozen=True)
4 class PolicyDecision1:
5     tenant_id: str
6     bundle_version: int
7     subject: str
8     action: str
9     resource_urn: str
10     allow: bool

typescriptOne Dark Pro

1 interface AuthzCtx1 {
2   tenantId: string;
3   bundleVersion: number;
4   subject: string;
5   action: string;
6   resourceUrn: string;
7 }
8  
9 export function cacheKey(c: AuthzCtx1): string {
10   return [c.tenantId, c.bundleVersion, c.subject, c.action, c.resourceUrn].join("|");
11 }

Operational notes

Note 1.1: Map Problem Statement & Authorization Control Plane metrics to allow/deny ratio and eval p99.

Note 1.2: Document bundle_version in JWT claims for Problem Statement & Authorization Control Plane cache coherency.

Note 1.3: Rate-limit policy admin APIs; require step-up MFA for Problem Statement & Authorization Control Plane mutations.

Note 1.4: Propagate binding revocations via event bus within 60s for Problem Statement & Authorization Control Plane.

Note 1.5: Shard policy tables by tenant_id to contain Problem Statement & Authorization Control Plane blast radius.

Note 1.6: Run chaos tests: PDP pod kill during Problem Statement & Authorization Control Plane peak traffic.

Note 1.7: Precompute role closures on publish to accelerate Problem Statement & Authorization Control Plane checks.

Note 1.8: Export sampled explain logs for Problem Statement & Authorization Control Plane SOC2 evidence.

Note 1.9: Validate SoD graph before activating bundle affecting Problem Statement & Authorization Control Plane.

Note 1.10: Use negative caching with 30s TTL for repeated Problem Statement & Authorization Control Plane denies.

Note 1.11: Align ReBAC tuple churn with Problem Statement & Authorization Control Plane cache invalidation messages.

Note 1.12: Canary-publish bundles to 1% tenants before Problem Statement & Authorization Control Plane global rollout.

If challenged on Problem Statement & Authorization Control Plane, answer with concrete numbers (QPS, p99, bundle propagation SLA) and deny-first semantics—not a flat "we use RBAC" statement.

Why interviewers care

Access Control System interviews reward crisp scope, explicit trade-offs, and failure stories—not generic microservice diagrams.

Interview checkpoint

Name one failure story for Problem Statement & Authorization Control Plane that proves you understand real outages, not happy-path diagrams.

Key Highlights

•AuthZ answers what a principal may do on which resource instance
•PEP at edge enforces; PDP evaluates versioned policy bundles
•Deny-first evaluation prevents privilege escalation by default

1	public final class AuthzCtx1 {
2	private final String tenantId;
3	private final long bundleVersion;
4	private final String resourceUrn;
5	public boolean isStale(long live) { return bundleVersion < live; }
6	public String cacheKey(String subject, String action) {
7	return tenantId + "\|" + bundleVersion + "\|" + subject + "\|" + action + "\|" + resourceUrn.hashCode();
8	}
9	}

1	from dataclasses import dataclass
2
3	@dataclass(frozen=True)
4	class PolicyDecision1:
5	tenant_id: str
6	bundle_version: int
7	subject: str
8	action: str
9	resource_urn: str
10	allow: bool

1	interface AuthzCtx1 {
2	tenantId: string;
3	bundleVersion: number;
4	subject: string;
5	action: string;
6	resourceUrn: string;
7	}
8
9	export function cacheKey(c: AuthzCtx1): string {
10	return [c.tenantId, c.bundleVersion, c.subject, c.action, c.resourceUrn].join("\|");
11	}