Skip to content
CSA Loom — the Microsoft Fabric experience for Azure tenants where Fabric isn't yet available: lakehouses, warehouses, notebooks, semantic models, Activator rules, Data Agents, across Commercial, GCC, GCC-High, and DoD IL5

fiab-0008: Deployment shape — two-tier (azd + Deploy-to-Azure); Marketplace deferred

Status: Accepted Date: 2026-05-22 Locked decision ref: LD-4

Context

The PRD originally specified a three-tier deployment surface: 1. Azure Marketplace Managed Application — one-click purchase + install 2. azd CLI — power-user path 3. Deploy-to-Azure template button — portal-click path

After the 2026-05-22 walkthrough, the user opted to defer the Marketplace surface and pricing model entirely to backlog (per OQ-10 + OQ-4 final answers). The reasoning:

  • Pricing model decision requires real-world adoption data Loom doesn't have in v1
  • Marketplace publishing flow (Partner Center + certifications + per- cloud Entra setup + publisher access agreements) adds ~6 weeks of v1 engineering that doesn't deliver customer value if pricing isn't decided
  • Federal customers in early adoption will run azd up against their own subs anyway (customer-managed deploy is the federal default regardless of Marketplace plan choice)
  • Once Loom has proven value + adoption velocity, pricing + Marketplace publish becomes a tractable next-step decision

Decision

Two-tier deployment shape for v1:

  1. azd up CLI — power-user path; full Bicep visibility; azd init -t fiab && azd up
  2. "Deploy to Azure" template button — embedded in README; opens Azure portal with pre-rendered ARM template; evaluator-friendly

Both deploy into the customer's own Azure subscription. Customer pays only for Azure consumption underneath. Loom IP (Console, Setup Wizard, parity services, Copilot runtime, docs, workshops) is free in v1.

Marketplace Managed Application + pricing model are deferred to backlog: - PRP-10 (Marketplace Managed App package) → backlog - PRP-11 reduced scope: per-boundary deploy validation workflows only (no Marketplace package validation, no Partner Center publish workflows) - §07.10 first-install timeline unchanged: 60-100 min from "begin" to working Console URL

Backlog items revisited after Loom v1 + v1.1 mature: - Pricing model (flat fee per capacity SKU + metered overage per DLZ is the leading candidate from PRD §07.9 but not locked) - Marketplace Managed App publishing (Commercial + Gov listings) - Publisher-managed vs customer-managed plan choice - Update mechanism via MCP-as-update-channel pattern

Consequences

Positive

  • v1 ships ~6 weeks faster by removing Marketplace publishing engineering
  • No Partner Center / Microsoft federal engagement required for IL5 in v1 (deferred to v1.1 with the IL5 boundary itself)
  • No pricing-model lock-in before adoption data is available
  • Customer-managed deploy is the default; matches federal preference (no publisher persistent access into Gov subs)
  • Loom remains free in v1 — zero procurement friction beyond the customer's existing Azure agreement

Negative

  • No Marketplace discoverability in v1 — Loom needs marketing pull (PRP-21) to reach customers
  • No publisher-push update channel — customers must azd up again or Console-pull updates via MCP (still works, but more manual)
  • Some federal customers prefer Marketplace procurement for budgetary / compliance reasons (Marketplace purchases hit MACC commitments differently than direct Azure consumption); Loom can't offer that v1
  • Loses the "AppSource" / "Microsoft Marketplace" SEO + discovery channel

Neutral

  • The Bicep platform underneath is identical to what a Managed App package would invoke — so v1.1 Marketplace publish is a wrapper exercise, not a rewrite
  • All Console + Setup Wizard + parity services design is unchanged from the three-tier vision — they're deployed by azd up instead of by Managed App, but they run identically post-deploy

Alternatives considered

Alternative Why not chosen
Three-tier (Marketplace + azd + Deploy-to-Azure) in v1 +6 weeks engineering; locks pricing model before adoption data
Marketplace-only Federal customers reject publisher-managed; ~6 weeks engineering for a path they'd then opt out of
Single-tier (azd only) Loses the evaluator-friendly Deploy-to-Azure button path; higher friction for first-time evaluators
Skip Marketplace work entirely (never plan a Managed App) Forecloses on the long-term SaaS distribution path; not the user's preference (defer, not skip)

References