Workload Identity, SPIFFE/SPIRE, mTLS and Zero-Trust Service Identity — DedSec Project
Workload Identity, SPIFFE/SPIRE, mTLS and Zero-Trust Service Identity Modern infrastructure increasingly replaces static service passwords with short-lived workload identities. This module…
Modern infrastructure increasingly replaces static service passwords with short-lived workload identities. This module explains identity issuance, attestation, trust domains, certificate/token rotation, service-to-service policy, and failure modes.
Learning objectives
By the end of this module you should be able to:
- Explain Workload identity and connect it to a concrete trust boundary or security invariant.
- Explain SPIFFE IDs and connect it to a concrete trust boundary or security invariant.
- Explain Attestation and connect it to a concrete trust boundary or security invariant.
- Explain Short-lived credentials and connect it to a concrete trust boundary or security invariant.
- Explain mTLS and connect it to a concrete trust boundary or security invariant.
- Explain Trust domains and connect it to a concrete trust boundary or security invariant.
Deep concepts
1. Workload identity
A workload identity represents a running service/process instance rather than a human. It should be derived from trusted workload/platform attributes and have a lifecycle aligned with the workload.
2. SPIFFE IDs
SPIFFE defines identity names within trust domains. The ID is an identifier, not authorization itself; policy decides what that identity may access.
3. Attestation
Identity issuance depends on node/workload attestation. The issuer needs trustworthy signals about where and what is running before minting credentials.
4. Short-lived credentials
Short-lived X.509 SVIDs or JWT-style credentials reduce dependence on manual secret rotation. Availability and clock/time correctness become important operational dependencies.
5. mTLS
Mutual TLS can authenticate both ends of a connection. Authorization still needs service/resource context, and certificate trust must be scoped so unrelated trust domains are not accepted accidentally.
6. Trust domains
Federating trust domains creates explicit cross-domain identity relationships. Keep mappings narrow and avoid translating broad external identities into powerful local ones.
7. Rotation and revocation
Short lifetimes can reduce the need for immediate revocation but do not eliminate emergency response. Plan issuer/key rotation, trust-bundle updates, compromised-node handling, and stale workload cleanup.
8. Policy and telemetry
Log workload identity, destination service, authorization decision, policy version, and certificate/token metadata without storing private keys. This makes service-to-service authority paths auditable.
Authorized exercises
Exercise 1 — Design a three-service toy architecture using short-lived workload identities and write an allow matrix for service-to-service calls
Design a three-service toy architecture using short-lived workload identities and write an allow matrix for service-to-service calls.
Exercise 2 — Model what changes when one node is considered untrusted: which credentials expire, what should be denied, and what evidence is needed
Model what changes when one node is considered untrusted: which credentials expire, what should be denied, and what evidence is needed.
Exercise 3 — Compare static API keys, cloud workload federation, and SPIFFE-style identities across rotation, attribution, and blast radius
Compare static API keys, cloud workload federation, and SPIFFE-style identities across rotation, attribution, and blast radius.
Knowledge checks
- Explain the most important trust boundary in Workload Identity, SPIFFE/SPIRE, mTLS and Zero-Trust Service Identity without naming a security tool.
- Give one example where cryptographic or authentication success still does not imply authorization success.
- Describe one stale-state or replay condition relevant to this module.
- Identify one telemetry source that can support a strong conclusion and one blind spot it still leaves.
- Explain how you would turn a discovered weakness into a regression test rather than a one-time finding.