What's in the Lab

Inventory · Azure Local Reference Topologies

See Sovereign Cloud Scenarios in Action

The ACX Lab runs a curated inventory of Azure Local deployment topologies on real hardware — not slideware. Every configuration below is racked, cabled, validated, and reservable today. Use it to demo customer scenarios, validate Microsoft Sovereign Cloud solutions, and let Field and Sales Engineers show prospects exactly what "supported and working" looks like.

Reserve a Topology → Connect to a Cluster

Every topology below maps to a published Azure Local reference architecture, so what a customer sees in this lab is exactly what they can order and deploy in their own environment — same validated hardware patterns, same network intents, same operational guardrails.

Azure Local Deployment Topologies in Inventory

🏭

Rack Aware Cluster Sovereign / Factory

A single cluster split across two physical rooms or racks (up to 4+4 nodes), each acting as its own local availability zone with ≤1ms RDMA latency between them. Built for factory floors, regulated facilities, and any site where hardware must be physically isolated for safety or compliance — while still behaving as one cluster to the operator.
Reference architecture ↗

🔌

Storage Switchless (2–4 Node) Edge / ADLess

Direct-connect storage between nodes — no TOR switch required for East-West storage traffic. The lowest-footprint, lowest-cost pattern for edge and disconnected sites, ideal for demonstrating "no dedicated switch fabric needed" scenarios to customers with constrained racks or remote locations.
2-node reference ↗ · 3-node reference ↗

🖧

Storage Switched, Converged Standard Datacenter

Dual TOR switch fabric with converged management, compute, and storage intents on a shared set of adapters. The default enterprise datacenter pattern — highest node-count scale (up to 16 nodes) and the configuration most customers deploying at scale will end up running.
Reference architecture ↗

🧩

Storage Switched, Non-Converged High Throughput

Dedicated adapters per traffic intent (management, compute, and storage each isolated) behind dual TOR switches. Used when workloads need guaranteed storage bandwidth independent of compute/management traffic — the pattern to show for latency-sensitive or high-IOPS customer workloads.
Reference architecture ↗

📦

Single-Node Micro-Footprint

One server, no dedicated storage fabric — the smallest possible Azure Local footprint. Great for retail, branch office, and space-constrained edge demos where the story is "this fits in a closet and still gets Azure Arc management, AKS, and VM hosting."
Reference architecture ↗

🎮

GPU-Accelerated Nodes AI / ML

NVIDIA A2 and A16 GPU-enabled nodes layered onto the standard cluster patterns above, for AI inferencing, computer vision, and GPU-partitioned VM demos entirely on Azure Local — no round-trip to the public cloud required.

Why This Matters for the Sovereign Cloud Story

Microsoft Sovereign Cloud customers ask the same three questions in almost every conversation: "Can this run fully isolated? Can it survive a rack going dark? Can I see it, not just read about it?" This lab answers all three on real, currently-racked hardware:

• Data residency & isolation — Rack Aware clusters demonstrate physical fault-domain separation without giving up single-pane-of-glass management, directly mapping to sovereign and regulated-industry data locality requirements.
• Disconnected & constrained-network operation — Switchless topologies show the minimum viable footprint for edge and low-connectivity sites, a recurring ask from public sector and defense-adjacent customers.
• Live, not simulated — Every cluster listed here is deployed, validated against the published Microsoft reference architecture, and reservable through Request Resources — so a Field or Sales Engineer can put a real customer scenario in front of a real prospect instead of a deck.

Bring a Customer or Prospect Into the Lab

Audience	Recommended Path
PM / Engineering	Reserve a topology to validate a new scenario or feature against real hardware before it ships. Start with Request Resources.
Field / Sales Engineering	Book a walkthrough session on the Rack Aware or Switchless cluster that matches your customer's industry vertical. Contact the ACX Lab Teams channel to schedule.
Customer Briefing	Use the live cluster as a backdrop for Sovereign Cloud conversations — see it running, not a static diagram.

Inventory reflects the current lab build; topology availability may shift with hardware refresh cycles. Check Lab Status before scheduling a demo.