{"id":950,"date":"2026-11-14T22:00:00","date_gmt":"2026-11-14T22:00:00","guid":{"rendered":"https:\/\/heartcareforyou.in\/blog\/?p=950"},"modified":"2026-11-15T17:00:00","modified_gmt":"2026-11-15T17:00:00","slug":"iamops-role-attribute-relationship-compare","status":"publish","type":"post","link":"https:\/\/heartcareforyou.in\/blog\/iamops-role-attribute-relationship-compare\/","title":{"rendered":"IAMOps Compared: Role, Attribute and Relationship Access Models"},"content":{"rendered":"

Anyone who has maintained a health-information website or an internal portal eventually faces the same design question: how should the system decide who sees what? Three families of answer dominate the literature, and each one makes IAMOps<\/strong> work feel different in daily practice. The comparison below lines them up using only the sources and standards the Ops Atlas cites, and it stays on the administrative technology side. Nothing here concerns diagnosis, treatment or any patient.<\/p>\n

The frame: what a decision needs<\/h2>\n

The Atlas, in its IAMOps entry<\/a>, reduces every authorisation decision to a requester, an action, a target and some context, then adds a per-platform rule for combining policies. That framing is useful because the three models differ mainly in where they look for the answer: at the requester’s job, at properties of everything involved, or at the links between things.<\/p>\n

Option one: role-based access<\/h2>\n

Role-based access attaches permissions to a named job, and people receive the job. A formal standard from 2004 (ANSI\/INCITS 359) gives the model a shared vocabulary, which partly explains its long life in organisations with clear job titles.<\/p>\n

Strengths.<\/strong> It is easy to explain to auditors, easy to teach to new staff, and easy to review: you read a short list of roles instead of thousands of individual assignments.<\/p>\n

Weaknesses.<\/strong> Roles tend to multiply. A team needs a small variation, someone clones a role, and a few years later nobody can say how three near-identical roles differ. The Atlas’s usage figures show the cost: roles often use only a small share of the actions they carry, and cleanup is slow because every role has an owner nobody has met.<\/p>\n

Option two: attribute-based access<\/h2>\n

Attribute-based access decides using properties: of the requester, the resource, the action and the environment. A US government publication from 2014, NIST SP 800-162, is the reference most often cited for the model.<\/p>\n

Strengths.<\/strong> A single rule can cover many cases (“people in the same department may edit documents carrying that department’s label”), so the rulebook stays short even as the organisation grows.<\/p>\n

Weaknesses.<\/strong> Rules depend on attributes being present and correct. The Atlas draws attention to what happens when a context value is missing: some comparisons fail safely, while negated ones can succeed. The model also demands good data hygiene, since a mislabelled document silently receives the wrong treatment.<\/p>\n

Option three: relationship-based access<\/h2>\n

Relationship-based access answers questions through links in a graph: a given person sits in a team, and that team owns a folder. The Atlas points to a 2019 paper from Google describing its global authorisation system, reporting latency under ten milliseconds at the ninety-fifth percentile, and to open projects inspired by it, including OpenFGA, which the Atlas lists as incubating in the CNCF.<\/p>\n

Strengths.<\/strong> It fits sharing-heavy products where permissions follow structure, such as folders, teams and delegations. Changes to a team update every dependent decision at once.<\/p>\n

Weaknesses.<\/strong> It introduces a service that must stay available and consistent. It is also a less familiar mental model for auditors, so evidence takes more effort to produce.<\/p>\n

A side-by-side summary<\/h2>\n