Hi Exchange Team! Looking for some clarification on supported models. This has come up in several design discussions on our end, and I’d like to make sure we’re interpreting the intended support boundaries correctly.
- For directory-synced users, if the goal is to have Exchange Online be the authoring plane for Exchange recipient attributes (email addresses / recipient properties) in the “cloud-managed remote mailbox” model described here, is setting IsExchangeCloudManaged = true required for that mailbox?
- If IsExchangeCloudManaged is not enabled (set to false), is Exchange writeback by itself intended to be sufficient to establish Exchange Online as the authoritative system for Exchange recipient attributes — or is writeback’s role limited to reflecting derived or normalized Exchange Online state (for example, attributes set or adjusted as part of mailbox provisioning or normalization) back to the directory, rather than serving as the primary authoring mechanism?
- For directory-synced users, are the supported models for ongoing Exchange recipient attribute management strictly limited to the following two patterns:
- AD-authored recipients, where IsExchangeCloudManaged = false and Exchange recipient attributes are authored and managed in on-prem AD
- Cloud-managed recipients, where IsExchangeCloudManaged = true and Exchange recipient attributes are authored and managed in Exchange Online
- …or is there a third supported model where IsExchangeCloudManaged = false but Exchange Online is still expected to be the ongoing authoring plane for recipient attributes (with writeback used to converge state), even though source-of-authority has not been explicitly transferred?
- In the AD-authored recipient model above (IsExchangeCloudManaged = false), if an organization intentionally does not use on-prem Exchange tooling and instead manually stamps Exchange recipient attributes directly in on-prem AD, is that considered a supported and expected configuration?
Thanks! Just trying to make sure we’re aligning to the intended support boundaries.
Also, sorry for such complicated questions. There has been alot of debate on our end as to what would work, and these are the points that seem to present the most confusion.