Friday, July 20, 2018

SD-WAN Deep Dive Part 2: Design and Implementation

July 2018 · Estimated Reading Time: 12 minutes

This is the second part of our deep dive series on SD-WAN. If you missed Part 1, where we covered overlay models, hardware footprints, and operational architectures, you can read it here. In this post, we shift our focus from architecture to implementation.

Routing Strategy and Policy Design

Modern SD-WAN solutions replace static route tables with dynamic, policy-based routing engines. Enterprises define application-driven policies—by DSCP, port, or even packet signatures—allowing real-time steering across underlay links. Some controllers allow nested policies that cascade across edge sites, enabling location-aware routing decisions.

QoS and Traffic Classification

SD-WAN vendors implement built-in QoS engines. They offer packet inspection, flow tracking, and bandwidth shaping. Traffic classification integrates with business policies, identifying mission-critical flows (like VoIP or ERP) and guaranteeing their performance. Marking packets at the edge and preserving DSCP across tunnels ensures end-to-end integrity.

Failover and High Availability Design

Failover mechanisms rely on link probing, jitter analysis, and SLA monitoring. Architectures now default to active-active link usage with seamless failover, using loss/jitter thresholds to trigger flow redirection. Hybrid setups (fiber + LTE) also rise as backup options. Multi-edge redundancy is handled via edge clustering or standby appliances.

Internet Breakout Models

Breakout design is a hot topic. Enterprises balance centralized vs distributed internet access. DIA (Direct Internet Access) at branches reduces latency for SaaS apps, but brings security concerns. Most deployments implement secure DIA using cloud-based SWG (Secure Web Gateway) or firewall-as-a-service (FWaaS) partners.

Security Policy Enforcement

Edge-to-edge tunnels provide encryption, but policy enforcement varies. Integrated NGFWs or service chaining to third-party firewalls (e.g., Palo Alto, Zscaler) helps bridge the security gap. More vendors embed URL filtering, malware protection, and DNS enforcement natively at the edge.

Orchestration and Change Management

SD-WAN orchestration platforms provide centralized push-based configuration, often via GUI or API. Policy rollouts include pre-checks, versioning, and staged rollouts. Some even allow intent-based change validation using digital twins or simulation. This minimizes outage risk during policy updates.

Lessons from Field Deployments

We see common implementation challenges: misaligned SLA thresholds, overzealous application definitions, and controller overload during failovers. Best practices include building test topologies, tuning telemetry thresholds, and incrementally introducing breakout policies with failback options.

Transition to Part 3

In our upcoming Part 3, we’ll dive into monitoring and optimization. Expect coverage on telemetry frameworks, anomaly detection, analytics, and ongoing tuning strategies.

 
Want help designing or troubleshooting your SD-WAN rollout? Reach out today. 
 


Eduardo Wnorowski is a network infrastructure consultant and Director.
With over 23 years of experience in IT and consulting, he helps organizations maintain stable and secure environments through proactive auditing, optimization, and strategic guidance.
LinkedIn Profile

Sunday, July 1, 2018

Revisiting MPLS TE in 2018: Viability, Use Cases, and Modern Alternatives

July 2018 • 7 min read

Understanding MPLS Traffic Engineering

Multiprotocol Label Switching (MPLS) Traffic Engineering (TE) emerged as a core mechanism in the early 2000s to optimize path utilization and route traffic based on constraints like bandwidth, delay, and administrative preference. Carriers adopted MPLS TE extensively to overcome limitations in traditional IGP-based shortest path routing.

TE Viability in 2018

As of 2018, MPLS TE remains a viable solution, especially in legacy environments where hardware investments and operational models still rely on RSVP-TE. In such settings, path predictability and granular control remain key requirements. However, several challenges persist:

  • RSVP-TE complexity in maintaining soft state and scalability across large backbones
  • Operational overhead in provisioning and adjusting tunnels
  • Limited interoperability across multi-vendor deployments

Segment Routing as a Disruptor

Segment Routing (SR) begins disrupting traditional TE approaches by enabling source-based routing and reducing control plane overhead. By encoding path instructions in packet headers, SR eliminates the need for per-flow state in the core. Combined with centralized SDN controllers, SR offers scalable and dynamic TE.

Comparing Use Cases

MPLS TE and SR address similar problems—optimal path selection, SLA enforcement, and failure recovery—but they differ in execution. In 2018, use cases for MPLS TE still dominate in networks with deep legacy investments or where control plane change is slow. Meanwhile, SR sees adoption in greenfield deployments and SDN pilots.

Operational Considerations

Network teams face a critical decision: continue maintaining RSVP-TE or begin transitioning to SR. Migration strategies include hybrid models, where RSVP-TE coexists with SR-TE to gradually phase out older mechanisms. Operators also explore intent-based networking where path constraints derive from policy rather than CLI configuration.

Vendor and Standards Landscape

Cisco, Juniper, and Nokia all offer robust MPLS TE and SR implementations. IETF support for SRv6, Path Computation Elements (PCE), and telemetry enhancements continues to strengthen the SR roadmap. TE in 2018 is no longer about whether to do it—but how to do it with less friction and more intelligence.

Final Thoughts

MPLS TE has served the industry well for nearly two decades. Yet, with SDN maturity and Segment Routing momentum, traditional TE sees diminishing returns. As 2018 progresses, network architects must evaluate when and how to shift toward simpler, more scalable TE architectures that align with evolving business needs.


Eduardo Wnorowski is a network infrastructure consultant and Director.
With over 23 years of experience in IT and consulting, he helps organizations maintain stable and secure environments through proactive auditing, optimization, and strategic guidance.
LinkedIn Profile

AI-Augmented Network Management: Architecture Shifts in 2025

August, 2025 · 9 min read As enterprises grapple with increasingly complex network topologies and operational environments, 2025 mar...