🔀 SD-WAN

Application-Aware Routing, Dual-Path Resilience & Cloud-Ready WAN

SD-WAN (Software-Defined Wide Area Network) replaces static, router-by-router configs with centralized policy and application-aware path selection. It steers each flow over the best available underlay (fiber DIA, MPLS, fixed wireless, LTE/5G, satellite) based on loss, latency, jitter, and business intent—with sub-second failover and deep observability.

Where SD-WAN fits in the SolveForce model:
🌐 Connectivity (Grammar)Connectivity • 🖧 FabricNetworks & Data Centers
☁️ Cloud (Syntax)Cloud • 🔒 Security (Semantics)Cybersecurity • 🧠 Decision LayerSolveForce AI


🎯 Outcomes (What SD-WAN Delivers)

  • Dual-/multi-path resilience — instant steer/failover across fiber + wireless + 5G + MPLS.
  • Better app experience — per-application SLOs (loss/latency/jitter) with brownout detection.
  • Cloud-ready edges — local Internet breakout to SaaS/IaaS with policy, or hub-and-spoke if required.
  • Faster changes — push policies from a controller; zero-touch provision (ZTP) new sites in minutes.
  • Proof — per-app dashboards, SLO compliance, change/audit trails, and carrier ticket evidence.

🧭 When to Use SD-WAN (and When Not)

Use SD-WAN when you need:

  • Active-active paths (fiber + fixed wireless/5G/satellite) and loss-aware steering.
  • Local Internet breakout for SaaS (M365, Salesforce) while keeping governance.
  • Cloud on-ramps (AWS Direct Connect / Azure ExpressRoute / Google Interconnect) with policy control. → Direct Connect
  • Rapid scale — dozens/hundreds of sites with consistent config and ZTP.

Pair with / Consider

  • MPLS (Multiprotocol Label Switching) where regulated or L3VPN contracts are required, often as an underlay alongside Internet. → MPLS
  • SASE (Secure Access Service Edge) if you want the security stack (SWG, CASB, FWaaS, ZTNA) delivered from the cloud next to SD-WAN. → SASE

🧱 Architecture (The Pieces)

  • Controller / Orchestrator — central brain for policy, inventory, ZTP, and upgrades.
  • SD-WAN Edges — CPE at branches/DCs/cloud (virtual or physical).
  • UnderlaysFiber DIAFiber Internet, MPLSMPLS, Fixed WirelessFixed Wireless, LTE/5GMobile Connectivity, SatelliteSatellite Internet.
  • Service Chains — route a flow through NGFW, IDS/IPS, DLP, or ZTNA (on-box or cloud security). → Cybersecurity

Common topologies

  • Hub-and-Spoke (simple, central services)
  • Partial/Fully Mesh (low-latency site-to-site)
  • Regional Hubs (cloud/on-ramp locality)
  • Cloud Edge (vEdge in VPC/VNet with BGP to TGW/ER/Cloud Router) → Cloud

🧠 Policy Model (Intent → Action)

  • App ID / DPI (Deep Packet Inspection) — recognize apps (SaaS/IaaS/VoIP) even if ports change.
  • Per-App SLOs — e.g., “Teams: loss≤0.1%, jitter≤15% latency, latency≤80 ms.”
  • Brownout vs. Blackout — detect degradation (brownout) and shift before a hard down (blackout).
  • Cost/Path Bias — prefer low-cost Internet until SLA breach, then escalate to MPLS/secondary.
  • QoS / Queues — prioritize voice/real-time; police bulk/backups to off-hours.
  • Path Conditioning — FEC (Forward Error Correction), packet duplication for voice, jitter buffers.

Policies are versioned and pushed; each change is auditable and roll-backable via the controller and ITSM.


📐 Transport Classes & Path Steering (SolveForce SLO Guardrails)

ClassTypical UnderlaysOne-Way LatencyJitter TargetLoss TargetNotes
AMetro fiber, wavelength≤ 2–5 ms≤ 15% latency< 0.1%DC/DCI, voice, trading
BRegional DIA/MPLS≤ 15–35 ms≤ 15%< 0.1%General enterprise
CContinental/global DIA (+ CDN/Anycast)≤ 80–120 ms≤ 15%< 0.1%Global SaaS/API
DLEO/GEO satellite, remotevariableengineeredengineeredRemote/backup

SD-WAN edges continuously measure these and steer per flow. Violations generate evidence for the NOC and—if carrier‐related—open tickets. → NOC ServicesCircuit Monitoring


🔒 Security Interlock (SD-WAN + SASE)

  • SASE = SD-WAN transport + cloud-delivered security (SWG, CASB, FWaaS, ZTNA).
  • ZTNA (Zero Trust Network Access) — per-app, per-session identity; replaces flat VPNs. → ZTNAZero Trust
  • Segmentation — VRFs per business unit; microsegmentation for crown-jewel apps in DC/cloud. → Microsegmentation
  • Crypto — IPsec tunnels/DTLS; MACsec on L2; TLS for SaaS with DLP/ATP. → Encryption

☁️ Cloud & On-Ramps (Design Patterns)

  • Local Internet Breakout — SaaS direct from branch; enforce SWG/DLP/SSL inspection via SASE.
  • Regional Hubs — break out near AWS/Azure/GCP regions for low-jitter SaaS/API. → Direct Connect
  • Cloud vEdges — deploy virtual edge in VPC/VNet; BGP to Transit Gateway / ER Gateway / Cloud Router.
  • Anycast Front Doors — publish the same VIP from multiple hubs for “closest healthy” entry. → BGP Management

🔭 Observability & Evidence (Prove It)

  • Per-app SLO dashboards — latency/jitter/loss by app & site; “good/brownout/blackout” status.
  • Underlay health — path loss, optical dBm, flaps, provider POP issues.
  • Overlay health — tunnel SLA, packet dup/FEC stats, QoE for voice/video.
  • Change audits — who changed what, when; success/rollback events.
  • Exports — logs/metrics to SIEM/observability for correlation and long-term proof. → SIEM / SOAR

🛠️ Integration Cheatsheet

  • Routing — BGP/OSPF redistribution; default-originate; PBR for edge cases. → BGP Management
  • DNS — split-horizon for SaaS; Anycast VIPs for APIs.
  • WAN Opt / Caching — only where needed; SD-WAN pathing usually beats legacy compression.
  • NTP/PTP — keep clocks sane for logs and voice; over-the-top GPS at hubs if required.
  • ZTP — ship edge, plug power/links; phone-home to controller; auto-join policy.

🧪 Reference Designs (By Outcome)

A) Resilient Branch (Voice + SaaS)

  • Underlays: Fiber DIA + Fixed Wireless; optional LTE/5G tertiary.
  • Policy: voice loss ≤ 0.1% → duplicate packets; Teams/Zoom jitter ≤ 15% latency; SaaS local breakout.
  • Security: SASE SWG + CASB; ZTNA for admin apps.

B) Cloud-First Enterprise

  • Hubs in colo with Direct Connect/ExpressRoute/Interconnect; branches steer cloud apps to nearest hub.
  • Anycast APIs; BGP communities mark “golden” routes; MACsec on L2. → ColocationDirect Connect

C) Remote / Harsh Links (LEO/GEO)

  • Policy favors FEC + jitter buffers; downgrades video to audio on sustained Class-D breach; store-and-forward for bulk.

🧾 Commercials & Licensing (What Drives Cost)

  • Edge count & bandwidth tiers; throughput licensing per device or pool.
  • Security bundle (SASE/SSE) add-ons (SWG, CASB, FWaaS, ZTNA).
  • Controller (cloud/SaaS vs. on-prem) and analytics retention.
  • Underlays — DIA/MPLS/wireless contracts, cross-connects in colo. → ColocationFiber Internet

✅ Implementation Checklist (No Surprises)

  1. Inventory & address plan — sites, subnets, IPv4/IPv6 overlaps; target SLOs by app.
  2. Underlays — primary fiber; secondary fixed wireless/5G; tertiary satellite if remote. → Fixed WirelessMobile ConnectivitySatellite Internet
  3. Policy — app catalog, per-app SLOs, cost/route bias, packet dup/FEC rules.
  4. Security — ZTNA groups, SWG categories, DLP rules, microsegments. → ZTNAMicrosegmentation
  5. Cloud — hubs and/or vEdges; on-ramp circuits; BGP policy. → Direct Connect
  6. ZTP & change — staging images, controller templates, maintenance windows.
  7. Observability — synthetics, packet loss maps, SIEM/SOAR hooks. → SIEM / SOAR
  8. Runbooks — brownout thresholds, carrier escalation, rollback, and RCAs. → NOC Services

🔄 Where SD-WAN Fits (Recursive View)

1) Grammar — controls flows across Connectivity underlays.
2) Syntax — optimizes paths to Cloud and data centers.
3) Semantics — enforces identity/inspection with Cybersecurity / SASE.
4) Pragmatics — signals drive SolveForce AI for prediction/auto-tuning.
5) Foundation — coherent terms under Primacy of Language.
6) Map — indexed across the SolveForce Codex & Knowledge Hub.


📞 Design an SD-WAN You Can Prove

Related pages:
ConnectivityNetworks & Data CentersCloudCybersecuritySASEZTNADirect ConnectBGP ManagementNOC ServicesCircuit MonitoringKnowledge Hub