Prefab, Fast-to-Deploy, AI-Ready & Audit-Grade
Modular Data Centers (MDCs) package power, cooling, racks, and network into factory-built modules—delivered, set, and energized in weeks instead of months.
SolveForce designs MDCs as a system: site prep → delivery → integrate power/cooling/fabric → Zero-Trust security → observability & evidence—with options for GPU/AI density, edge use, or temporary/DR capacity.
Related hubs: 🏢 On-Prem DCs → /on-prem-data-centers • 🧭 Edge DCs → /edge-data-centers • 🏢 Colocation → /colocation
🌈 DCI & On-Ramps → /wavelength • /lit-fiber • /dark-fiber • /direct-connect
🎯 Outcomes (Why Modular with SolveForce)
- Speed-to-serve — factory integration + parallel site prep = weeks to rack-ready.
- Predictable quality — repeatable, tested subsystems; fewer on-site surprises.
- Scalable blocks — add pods for compute, storage, or GPU capacity without re-architecting.
- Anywhere footprint — campus lots, rooftops, yards, remote/industrial sites, disaster zones.
- Audit-grade ops — as-builts, baselines, and SLO dashboards streamed to SIEM/SOAR.
🧭 Scope (What We Deliver & Integrate)
- Form factors — ISO/containerized pods, skid & enclosure builds, multi-module halls (power, cooling, IT).
- Power — utility tie-in or gensets; UPS (double-conversion), battery autonomy, A/B PDUs; busways or whips.
- Cooling — in-row/close-coupled DX/CRAH, rear-door heat exchangers, liquid-to-chip options for AI density.
- Racks & Cabling — pre-wired rails, latching PDUs, structured cabling (Cat6A/MPO), labeling & OTDR packs. → /structured-cabling • /racks-pdu
- Fabric — compact leaf/spine, 10/25/40/100/400G; EVPN/VXLAN; MACsec/L1 encryption where required. → /networks-and-data-centers • /encryption
- Backhaul — fiber laterals (lit/wave/dark), fixed wireless, LTE/5G, satellite; IPsec/MACsec/L1 policies. → /lit-fiber • /wavelength • /mobile-connectivity • /satellite-internet
- Compute/Storage — virtualization/K8s, SAN/NVMe, and GPU/AI nodes. → /kubernetes • /san • /bare-metal-gpu
- Security — physical (access control, CCTV), NAC/802.1X, microsegmentation, ZTNA/SASE for admins/users. → /nac • /microsegmentation • /ztna • /sase
- Continuity — snapshots, immutable backups, orchestrated DRaaS. → /cloud-backup • /backup-immutability • /draas
- Observability — DCIM (power/thermal/door/leak), fabric/storage/compute telemetry → NOC/SIEM. → /noc • /siem-soar
🧱 Building Blocks (Spelled Out)
- Power train — utility/genset → ATS → UPS (N/N+1/2N) → A/B PDUs → metered outlets; load-bank tests archived.
- Thermal design — containment (hot/cold), RDHx or liquid loop for high kW/rack; thermal maps & alarms.
- Envelope — wind/seismic/flood ratings, EMI shielding (where required), access control, tamper detection.
- Network — EVPN/VXLAN, Anycast gateways; QoS for storage/voice/control; OOB mgmt network.
- Compliance — HIPAA/PCI/CJIS/CMMC enclaves via cages/VRFs + microseg + key custody.
🛠️ Reference Patterns (Choose Your Fit)
A) Rapid Edge/MEC Pod
3–8 racks; UPS + in-row cooling; dual uplinks (fiber + LTE/5G); K8s cluster; ZTNA for admin; DR to core.
B) GPU/AI Modular Block
Liquid-ready racks (rear-door/LC-direct), IB/RoCE spine, NVMe scratch + parallel FS, wave backhaul; PUE & thermal SLOs. → /bare-metal-gpu
C) Campus Expansion Pod
Leaf/spine tied to campus core; SAN integration; Object-Lock backups; Anycast services; SD-WAN for branches.
D) Regulated Enclave in a Box
Cage + VRF/microseg; MACsec/IPsec; HSM/Managed HSM; immutable logs/backups; evidence packs for audits. → /key-management • /cybersecurity
E) Temporary / DR Staging Pod
Short-term capacity for events/outages/migrations; scripted cutover; failback runbooks; shipping-ready teardown.
📐 SLO Guardrails (Targets You Can Measure)
| KPI / SLO | Target (Recommended) |
|---|---|
| Factory-to-rack-ready (typical) | 4–10 weeks site-dependent |
| Power availability (A/B) | ≥ 99.99% at rack |
| Thermal headroom (p95 inlet) | Within ASHRAE envelope |
| Leaf↔Leaf latency (intra-pod p95) | ≤ 10–50 µs |
| DCI (metro one-way) | ≤ 1–2 ms (wave/EPL) |
| GPU/AI rack density | 20–80+ kW/rack liquid-ready |
| Evidence completeness | 100% (as-builts, OTDR, load-bank, RFC tests) |
SLO breaches trigger SOAR actions (reroute, shed load, thermal throttle, rollback) and incident tickets. → /siem-soar
🔒 Zero-Trust, Physical + Logical
- Physical: badge+biometric, mantrap (where feasible), CCTV retention, escorted access logs.
- Logical: 802.1X/NAC on ports, ZTNA for consoles, microseg for east-west, PAM for elevation, vault-managed secrets & HSM keys. → /pam • /secrets-management • /key-management
📊 DCIM & Telemetry (Evidence by Default)
- Power (A/B), PDU load, UPS/genset states; thermal sensors & RDHx loop health; door/tamper/leak.
- Fabric: latency/jitter/loss, light levels (if optical), BER/FEC; compute/storage SLOs; exported to NOC/SIEM with WORM options.
→ /noc • /circuit-monitoring • /siem-soar
💵 Commercials (What Drives Cost)
- Power density (kW/rack), liquid vs air cooling, number of racks/modules, optics/backhaul, DCIM, security layers, delivery logistics (crane/pad), permitting.
- Managed vs co-managed operations; maintenance windows & spares.
🛠️ Implementation Blueprint (No-Surprise Rollout)
1) Use cases & SLOs — latency, kW/rack, GPU needs, DR tiers, compliance tags.
2) Site due diligence — soil/pad, drainage/flood, wind/seismic, grounding/bonding, crane access, conduits.
3) Utilities — service entrance, ATS, UPS topology, genset sizing, fuel logistics.
4) Cooling — containment plan, RDHx/liquid loops, thermal acceptance tests.
5) Fabric & backhaul — EVPN/VXLAN, OOB; wave/lit/dark/fixed-wireless/LTE/5G; MACsec/IPsec/L1 policy.
6) Compute & storage — racks/tier layout, SAN/NVMe, GPU pods, CSI; snapshot/replication plan.
7) Security — NAC/802.1X, ZTNA/SASE, microseg, WAF for public edges; vault/HSM; SIEM/SOAR wiring.
8) Baselines & acceptance — load-bank, thermal step, OTDR, RFC 2544 / Y.1564 fabric tests; archive artifacts.
9) Operate — DCIM/NOC thresholds, SLO dashboards, patch & firmware windows, quarterly tune-ups.
✅ Pre-Engagement Checklist
- 📍 Site(s) & constraints (pad, permits, power feeds, grounding).
- ⚡ Target kW/rack & total kW; UPS autonomy; genset runtime.
- ❄️ Cooling approach (air, RDHx, liquid-to-chip); thermal envelope.
- 🧰 Rack count, PoE/low-voltage needs, cabling routes.
- 🖧 Backhaul mix (wave/lit/dark/fixed-wireless/LTE/5G/sat), diversity letters.
- ☸️ Platform choices (VM/K8s/bare metal), GPU density, storage tiers.
- 🔐 Identity (SSO/MFA), NAC posture, ZTNA/SASE policy, keys/secrets posture.
- 💾 Backup/DR tiers, Object-Lock scope; test-restore cadence.
- 📊 SIEM/NOC destinations; SLO dashboards; escalation tree; change windows.
- 💰 Budget guardrails; managed vs co-managed ops; delivery timeline.
🔄 Where Modular DCs Fit (Recursive View)
1) Grammar — packaged rails in /networks-and-data-centers & /connectivity.
2) Syntax — compose with /cloud and /colocation for hybrid/edge.
3) Semantics — /cybersecurity preserves truth (identity, crypto, segmentation, evidence).
4) Pragmatics — /solveforce-ai predicts power/thermal risk, backhaul constraints, and capacity.
5) Foundation — consistent terms via /primacy-of-language.
6) Map — indexed in the /solveforce-codex & /knowledge-hub.