Should SD-WAN replace MPLS?

January 29th, 2018 by · 2 Comments

This Industry Viewpoint was contributed by Steve Brar, Director of Solutions Marketing at Riverbed.

As SD-WAN gains more widespread adoption, you may be asking, “How should I compare SD-WAN vs MPLS?” This is an interesting question, but thinking about networks this way misses the central distinction between SD-WAN and MPLS. SD-WAN capabilities are a significant enhancement to enterprise networks, and work with a range of underlying technologies, including MPLS.

A Brief History of MPLS

A bit of history (skip to the next paragraph if you were working in networking 20 years ago). MPLS was invented at a time when complex networks were straining the capacity of the available processors for routers. Networking companies and their customers needed a switching protocol that could reliably deliver quality, security, and speed across wide-area links, with the capability to carry multiple protocols and quickly re-route traffic in the event of a link failure. Once it gained the acceptance of enough vendors and telecom providers, MPLS was broadly deployed, and was a good fit with the centralized data centers and hub-and-spoke network architectures of the time.

MPLS certainly remains a viable option for heavily used links between major sites, for redundant connections with fast failover, and for business-critical real-time applications that are hosted in the data center. However, as the reliability and service quality of other options have improved, organizations are increasingly using VPNs and lower-cost Internet broadband network links for many of their network nodes.

The Flexibility of SD-WAN

This is where the benefits of SD-WAN come into play. It is simply no longer feasible to spend hours of experienced network staff time adjusting hardware and configurations on individual routers to deal with frequent traffic changes -- not to mention the complexity of the command line interfaces and the potentially disastrous consequences when an error is made. SD-WAN is a modern technology  that brings the benefits of virtualization to networking, as has already happened with servers and storage.

SD-WAN introduces a better way to manage the dynamic nature of traffic between links, whether they are high-cost MPLS links or low-cost broadband connections. It enables organizations to reduce network costs and adjust the network design to the actual traffic flows between data centers, branches, and the cloud more effectively and efficiently.

A Few Scenarios

There are three common scenarios that can be used to illustrate the differences and benefits of an SD-WAN: hybrid WANs, cloud connectivity, and branch networking.

Hybrid WANs

Organizations with their own data centers or private clouds frequently find themselves with a hybrid network of MPLS links between major sites, and a growing set of other types of connectivity to smaller sites, partners, and other related entities. As desirable as it may be, MPLS links to every possible work location are simply not cost effective. Managing the complexity of these hybrid WANs with static router configurations results in inefficient traffic paths and poor quality of service for bursty and bandwidth-intensive applications.

Using SD-WAN, network administrators can create path-selection rules appropriate to the business, application, and user requirements. Traffic is routed to the appropriate path based on a variety of characteristics, including reliability, available bandwidth, and cost. Network capacity can ebb and flow as needed, without requiring any hands-on intervention. With this approach, applications that need the reliability and quality of service of MPLS can get it, and even have an Internet-based backup alternative, while all other traffic is sent by the best available route.

Cloud Connectivity

The increasing use of cloud services can put a significant strain on pure MPLS network architectures. All traffic goes to the hub first, and from there out to the Internet, creating bottlenecks for business-critical SaaS applications, or IaaS services such as Amazon AWS and Microsoft Azure. Continually increasing the bandwidth of the centralized Internet link, the speed of the routers that feed it, and the capacity of the security devices that protect it can consume a significant portion of networking team’s time and budget.

Instead, local Internet broadband connections and VPNs deliver the desired performance and security, at a fraction of the cost. SD-WAN management readily handles these configurations, protecting external traffic with embedded encryption and delivering much better performance for cloud services. Many organizations are finding that, with the majority of their traffic now destined for the Internet, this architecture enables them to reduce the capacity of some of their MPLS links, further reducing costs.

Branch Networking

Hub-and-spoke MPLS network architectures work great when most of the traffic is headed for the data center. However, as application and data movement becomes more varied, with an increasing amount of branch-to-branch and branch-to-Internet, the network suffers from “tromboning”, with traffic having to always cross through the hub even if it is going to a neighboring branch. This results in network delays, congestion, and increased latency.

SD-WAN makes it easy to manage mesh broadband VPN connections between sites, improving traffic flows and simplifying network management. Administrators can configure and update software-defined virtual routers remotely with zero touch, quickly and consistently apply business policies across all links, with no need for physical access to at the branch and secure communications with identity-based access and network segmentation.

Asking the Right Questions

The question is not SD-WAN vs MPLS. It is about selecting the appropriate network connections for the organization’s needs and using SD-WAN to easily deploy and effectively manage them. Choose the best combination of options for each site, whether that is data center or cloud, wired or wireless, private or Internet. Then create a unified fabric with centralized management that applies consistent polices across networks spanning a diverse set of locations and supporting hundreds of applications.

Orchestrate network utilization with automated SD-WAN workflows based on user, location, application, and security requirements, not static ports and IP addresses. Optimize network performance by dynamically switching between network paths based on real-time link performance characteristics, such as latency, packet loss, and jitter. Incorporate cloud services such as AWS and Azure on demand, without lengthy provisioning delays. Like virtualized compute and storage services, SD-WAN separates the physical network from the control plane, delivering the agility and cost savings that modern organizations need to be productive and competitive.

Steve Brar is the Director of Solutions Marketing for Riverbed. In this role, he leads the marketing strategy for Riverbed's Application Performance Platform and cross-portfolio solutions. Steve has been with Riverbed since 2014. Prior to joining Riverbed, Steve led product marketing for HP's campus networking product lines. At HP he held engineering, product management roles, and product marketing roles. He has more than 12 years of experience in the networking industry. Steve graduated with a BS in Computer Science & Engineering from the University of California, Davis. He is currently based in San Francisco, California.

Categories: Industry Viewpoint · SDN

Join the Discussion!

2 Comments So Far


  • Anonymous says:

    Should? it already is! … Old news.

    • Anonymous says:

      Not really….to be honest the Private Networks will never go away 100%. Hybrid Networking using a combination of MPLS/DIA/BB/LTE for a multipath delivery of traffic based on business requirements is a must.

Leave a Comment

You may Log In to post a comment, or fill in the form to post anonymously.





  • Ramblings’ Jobs

    Post a Job - Just $99/30days
  • Event Calendar