This is the repository for the development of IDQL Policy Language.
IDQL's objective is to standardize access policy and associated APIs across the cloud and the stack usable in standalone deployments all the way to hybrid multi-cloud scenarios.
Quick link to IDQL specifications.
IDQL's goal is intended to address the following issues:
- Multi-cloud access policy does not exist;
- Incumbent vendors are not motivated to address multi-cloud challenges in the absense of a standard;
- Distributed architectures require policy consistency across disparate platforms, domains and technologies; and,
- A standard will provide improved consistency, risk control while avoiding security silos, greater costs, and lock-in.
IDQL is intended to enable management of the following policy scenarios where policy may control:
- Which administrators may make changes in the environment;
- What service features, configurations, and even SKUs may be deployed;
- Used to define what applications may be deployed (for example, an admission controller in a Kubernetes cluster controls what application tags may be deployed in a cluster);
- Mandate network requirements for an application including restrictions for a software defined network (SDNET);
- Configure the authentication requirements for an Identity Provider;
- Operation of a platform service such as an Identity Aware Proxy;
- External access to application endpoints;
- Internal access between micro-services;
- Scoping of information available to an application; and,
- Inform applications what rights a subject has in order to optimize the user interface experience.
- CNCF Open Policy Agent (OPA) – Focused on K8S cluster management, networking, and microservices. It is expected that IDQL will be supported in OPA by using a set of (tentatively planned OPA Rego modules that will be able to interpret IDQL policy directly in OPA Agents.
- CNCF SPIFFE/Spire – focused on App to App identity using x509, not end user identity.
- SAML, OIDC, OAuth are all protocols for SSO and Authorization but not end user identity policy.
- XACML (OASIS) – focused on fine-grained entitlements not end user identity policy. Not declarative, requires custom code and is too complex.
IDQL Policy is a series of meta statements that define simple policy rules that are then translated and deployed to
the correct cloud providers, layers and components. A basic statement consists of a Subject + Action + target
Object + condition.
The IDQL policy specification defines a neutral JSON policy rule format. IDQL is a platform-neutral format that can be used to collect, analyze, audit, and provision policies. The format has been tested against 3 broad categories of policy systems including:
- Classic RBAC systems that use role based groupings to manage access (uses virtualized policy mapping)
- Declarative Policy Language systems that are mapped into IDQL. Examples include Google Bind, and Amazon Cedar.
- IDQL Native Policy decision systems such as implemented with the Open Policy Agent and the Hexa OPA Project.
We are working with developers and customers to build a set of specifications and open source as part of the Cloud Native Computing Foundation. The IDQL Working Group will be responsible for the policy specification published on this project.