Discovery Phase Tool: This builder walks through the key decisions required to specify a complete vendor-agnostic SMB datacenter solution — 3 servers, 2 switches, 1 shared storage array. Outputs are designed to start the conversation with your SE or VAR, not replace it.
Step 1 — Compute & Memory
Processor platform, core density, and memory architecture
Processor Platform
Intel Xeon
8 Memory Channels / Socket
Broadest ecosystem support
Strong ISV certifications
Familiar to most VAR channels
Slight edge on niche workloads
AMD EPYC
12 Memory Channels / Socket
More cores per dollar
Lower TDP — runs cooler
More memory bandwidth
Growing enterprise adoption
Intel Xeon holds the largest installed base and enjoys the broadest VAR and ISV support. For 99% of virtualized SMB workloads either platform executes x64 code identically — platform choice often comes down to vendor pricing and support preference.
Core & Socket Strategy
Single socket configuration limits PCIe lane availability. High card density requires a second CPU to activate all expansion slots.
Max Cluster VMs
--
N-1 failover @ 4:1 vCPU
Avg vRAM / VM
--
Virtual allocation ceiling
Physical RAM / Node
--
Balanced per architecture
Est. Node Power
--
CPU + memory overhead
Architect's Note: Physical RAM is your hard floor. Modern hypervisors use memory ballooning and transparent page sharing to stretch allocation — but sizing to physical is always the right starting point.
Step 2 — Boot Drives
Local storage for hypervisor only — workload data lives on shared storage
Boot Storage Option
Boot Storage Reference
BEST
Dedicated Boot Module (BOSS/NS204i) — Vendor-specific M.2 RAID card that keeps boot drives off PCIe and SATA backplane. Cleanest solution, no slot consumed, purpose-built for hypervisor boot.
GOOD
NVMe SSD Pair — Modern hypervisors require frequent reads and writes to boot media. NVMe provides the performance headroom. A mirrored pair (RAID 1) adds resilience. Two M.2 slots consumed.
OK
SATA SSD Pair (480GB) — Sufficient for boot-only workloads. Budget-friendly. Consumes SATA backplane ports. Adequate if NVMe slots are being used for other purposes.
Important: In a 3-2-1 architecture, local drives on each server are for hypervisor boot only. All VM storage, snapshots, and workload data reside on the shared storage array — not on local drives. This is intentional: it enables VM mobility (live migration) between nodes and centralizes data protection.
Boot Option
--
Per node
PCIe / Slots Used
--
For boot storage
Step 3 — Connectivity & Fabric
Production Ethernet, storage fabric, and out-of-band management
Ethernet Strategy
Copper vs. Fiber: 10Gb copper (RJ45) connects directly to standard switches and is backward-compatible with 1Gb infrastructure. Fiber (SFP28) offers higher density and better signal integrity over distance but requires SFP transceivers and fiber cabling — higher cost, higher performance.
Storage Fabric (Optional)
FC vs. iSCSI: Fibre Channel is a dedicated storage protocol — highest performance, lowest latency, highest cost. Most SMB deployments use iSCSI (storage over Ethernet) which eliminates the need for HBAs and separate FC switches. FC is worth the investment for latency-sensitive workloads like large SQL or Oracle databases.
Cluster Network Footprint (3 Nodes)
Switch Ports Required
--
Eth + OOB ports
PCIe Slots / Node
--
Expansion cards
Cluster Heat
--
BTU/hr total
Stack Rack Units
--
Servers + switches + storage
Out-of-Band Management: Dell iDRAC, HPE iLO, and Lenovo XCC each provide a dedicated 1Gb RJ45 management port built into the server motherboard. These connect to a separate management VLAN and do not consume any PCIe expansion slots.
Step 4 — Shared Storage Array
The "1" in 3-2-1 — centralized storage for all VM workloads
Array Form Factor
--
Rack space
Est. Array Power
--
Watts
Protocol alignment: If Fibre Channel is selected as storage protocol, HBA cards must be added in Step 3. iSCSI and NFS run over existing Ethernet infrastructure — no additional hardware required.
Step 5 — Support & Service
Contract term and response time level
Contract Term
Most Common
3-Year Term
Standard SMB contract. Aligns with typical refresh cycles. Usually the best price-per-year value. Renewable by vendor at expiration.
5-Year Term
Lower annual cost. Good for stable environments with predictable workloads. Less flexibility if technology needs change.
Response Level
Most Common
Next Business Day
Parts and technician on-site by next business day. Standard for most SMB environments. Assumes overnight is acceptable downtime window.
4-Hour Response
Mission-critical workloads. Technician on-site within 4 hours, 24x7x365. Meaningful cost premium — justified for revenue-generating systems.
Rails note: All 3 servers require rack rail kits. These are typically ordered separately and are vendor and rack-post specific (square hole vs. round hole). Confirm rack post type before ordering. Cable management arms (CMA) are optional but recommended for dense deployments.