Discussion: Approaches to multi-tenancy with EDVs

[tplooker]

On the 13th of May WG call a good discussion on the considerations that surround the different approaches to expressing multi-tenancy in URLs used for communication with an EDV server was had, brought about primarily due to a conversation on how best to achieve discovery see #12. Currently in the specification today the tenancy (essentially the EDV instance abstraction) is expressed through a URL path convention (e.g https://example.com//docs/), another common approach to multi-tenancy is to leverage sub-domains (e.g https://.example.com/docs/). However there are distinct tradeoffs that come with these different approaches, for example when using the subdomain approach EDV instances would be fire-walled from one and other when interacting via a User Agent (browser). Where as when using the path based URL convention, mechanisms like /.well-known endpoints cannot be used as a discovery tool for EDV instances.

[tplooker]

Discussed on the 24th June WG call, proposed resolution is for the edv spec to remain agnostic or permissive of both models for multi-tenancy. The spec should include language that documents this position.

[msporny]

Since EDV's use ZCAPs for authorization, multi-tenant systems can be constructed in a variety of different ways. The specification shouldn't prevent multi-tenant systems that are domain-based or path-based and should speak to why implementers shouldn't assume one or the other design pattern when implementing their EDVs.