matt-taylor.tech
← Back to projects

Work · National food brokerage

Enterprise naming and identity standards

Entra ID dynamic groups Active Directory Intune Conditional Access Microsoft 365 groups PrinterLogic IP address management

I designed the enterprise naming and identity standards that every downstream automation depends on: dynamic group membership, license assignment, Conditional Access scoping, Intune policy targeting, regional printer assignment, MDM compliance grouping, distribution-list maintenance, asset tracking, network segmentation, and acquisition integration.

These are not aspirational documents. They are the patterns the production tenant actually runs on, and the reason a new office or a newly acquired entity can be provisioned without inventing anything.

One attribute carries the weight

The foundational decision: a region code lives in the Entra ID Department attribute on every user object, drawn from a documented taxonomy of active codes plus retired ones that record prior structure. That single attribute is the input to every dynamic group rule, license assignment policy, Conditional Access scoping decision, and workflow downstream.

Concentrating the dependency that way is a deliberate trade. It makes provisioning nearly free, because setting one field correctly propagates everywhere. It also means getting that field wrong breaks automations across the whole organization, so the standard exists to make the correct value unambiguous.

Computer naming

Format: {REGION}-{TYPE}{SERIAL}. The design win is using the device's BIOS serial number rather than a sequential counter:

  • Intune auto-names machines at enrollment with no human in the loop and no per-device sequence to track.
  • Dynamic Entra ID groups resolve on a name prefix, so a laptop in a given region lands in the right group automatically.
  • Regional printer queues auto-deploy off the same prefix.

Two prior formats are explicitly documented as retired: a sequential scheme that required manual tracking and blocked automation outright, and a legacy numbers-only scheme carrying no region or type signal at all. Every standard names the format it replaced and why the old one failed, which is what makes the change defensible rather than just a preference.

Server naming

A composed format carrying region, location, platform, role, and instance number, with role codes spanning the full range of server functions. The name tells you what a box is and who it serves before you look anything up.

The design choices that make it survive contact with operations:

  • Region means the business region the server supports, not where it physically sits. A box housed in one region serving another carries the region it serves.
  • Virtual machines get a hypervisor-agnostic code, so a VM can move between hosts without a rename. Cloud instances carry a provider code.
  • The OS code is at segment level only. The exact version lives in asset management, so an in-place upgrade doesn't force a rename.

The through-line is that a name should only encode facts that don't change, and anything volatile belongs in the system of record instead. Every naming standard that ages badly does so because it encoded something that later moved.

Group naming, on-premises and cloud

The on-premises Active Directory standard covers the server-management and RDS-authentication scope: five purpose codes, computer-object groups for server collections, user-object access groups by region, and administrative tiers. Template accounts sit in a service-accounts OU under an underscore prefix so they sort to the top, and copying a template inherits its group memberships automatically.

The cloud standard is its counterpart, deliberately given a distinct prefix so an on-premises group is never mistaken for a cloud group in a directory search. It codifies thirteen purpose codes plus structured subsection schemas for the high-cardinality cases:

  • Conditional Access: paired include and exclude groups, so every policy has both and neither is improvised.
  • Intune: eight subsections spanning application deployment, configuration, remediation scripts, platform scripts, compliance, VPN, Wi-Fi, and certificates.
  • Power Platform and SharePoint, the latter with a permission suffix appended so the access level is legible from the name.

Assignment type (manual or dynamic) and object type (user or device) are baked into the name too, so the audience and the automation behavior are visible at a glance rather than requiring a portal lookup.

Directory groups built for humans

A separate taxonomy governs the company's mail-enabled groups, designed for end-user discoverability rather than admin readability. Every group name starts with a common token so they cluster in the address list, and ends with a common suffix so search-by-suffix works. Beneath that sit company-wide, business-unit, regional, role, department, and region-plus-department groups, along with city-and-state groups auto-populated from user attributes.

The standard also carries a documented exception list for a handful of addresses that use a concatenated form, because the natural form collided with an existing distribution list or shared mailbox. That kind of nuance is what separates a standard written for a brownfield tenant from one written for a greenfield diagram.

Addressing

An enterprise IP plan covers the full office estate on the same logic: per-region address blocks, purpose-coded segments so segment identity reads the same at every site, consistent role conventions within each segment so an engineer reading a firewall rule knows what a given host does without a lookup, a matched out-of-band management pattern, and office sizing tiers. Segment purpose and VLAN tag are explicitly decoupled, and the plan is on its third documented revision.

How they interlock

The seven standards are not independent documents. The region taxonomy is the input to all of them. Computer names feed dynamic groups, which feed Intune deployment. Server names live in AD groups, which authorize against session hosts addressed per the IP plan. Directory group membership rules read the same attributes the region taxonomy governs. And the acquisition integration playbooks extend all of it into new entities, so naming, identity, and addressing land consistently on day one rather than being retrofitted later.