Security model

Dyncloud's security is built in layers (defense in depth): if one measure fails, there's another behind it. No layer depends on the others being perfect.

Defense in depth

Layer	What it protects
Access	Strong identity, least privilege, access logging
Network	Segmentation, encryption in transit, minimal exposure
Data	Per-environment isolation, encryption in transit, separate backups
Secrets	Dedicated secrets manager, no plaintext credentials
Platform	Patching, vulnerability management, self-healing
Operation	Versioned, auditable, and reversible changes

Principles

Secure by default: protections come enabled; there's nothing to remember to turn on.
Least privilege: each person and each system has only the permissions it needs.
Minimal exposure: only what's essential is published to the Internet; the rest stays on segmented internal networks.
No implicit trust: components don't trust each other just for being on the same network; communication is restricted to what's necessary.
Traceability: administrative access and changes are logged.

The layers in detail

Access control — who gets in and with what permissions.
Encryption — protection of data in transit and secrets at rest.
Secrets management — credentials and keys.
Network security — segmentation and exposure.
Vulnerability management — patching and response.

Philosophy

We don't chase security as a box to tick, but as a property of the design: the platform is secure because it's built that way, not because security was bolted on at the end.