SD-WAN: Right for the Cloud Era?
Not only has WAN transformation been talked about for decades, but SD-WAN in particular has been a red-hot topic for the last couple of years. Investors seem to believe the market will stay this way as well; over the past couple of months both VeloCloud and Aryaka have raised additional funds to be able to meet the explosion in user demand for SD-WAN.
One of the core tenets of my research is that the best opportunity to gain customer share is when markets undergo transitions, which is why we see so much startup activity in this market right now.
Why Is the WAN in Transition?
That's an easy answer. Legacy WANs are broken and have been for decades. Prior to becoming an analyst I was in corporate IT, and back in the '90s we discussed WAN transformation. However, unlike today, there really wasn't a viable alternative at the time. Also, traditional WANs were inflexible, inefficient, and overly expensive. It wasn't holding the business back so most IT departments took an "if it ain't broke, don't fix it" attitude regarding the wide area network.
Today, the world is quite different. The cloud has completely changed WAN traffic patterns so legacy WANs are indeed having an impact on worker performance. Understanding why requires some knowledge of different compute models.
A decade ago client server computing was dominant. With this model, data is stored in the company data center and then accessed by workers in branch offices. A legacy network uses a hub-and-spoke design where the data center is the hub and branch offices are connected via the spokes. So from this perspective, the network architecture was actually optimal. The problem was that managing the network was done one box at a time, and it could take forever to make even simple changes.
Additionally, hub and spoke is sub-optimal for the Internet as traffic flows from the Internet into the hub location and then to the branch via one of the spokes. The traffic will then "trombone" at the branch, and then head back down the same spoke, back through the hub, and then out to the Internet. With this architecture, Internet traffic is passed over the company WAN twice, using much more bandwidth than is necessary. During the client server era, Internet traffic was best effort in nature and made up only a small percentage of network traffic, so businesses put up with the inefficiencies of this model.
Fast-forward a decade and the use of cloud has exploded. CRM, ERP, UC, contact center, productivity apps, storage, servers -- you name it, it's available in the cloud. Many organizations today have a mandate to be cloud first whenever possible, driving Internet traffic to all-time highs. The trombone effect can negatively impact application performance, causing companies to need to upgrade their WANs -- a very expensive task if MPLS is the primary technology in place.
SD-WAN to the Rescue
The SD-WAN can be a panacea for the network by lowering costs and improving user experience as it brings an entirely new architecture.
With an SD-WAN, a high-price MPLS network can be augmented or replaced with broadband, which is much lower cost. However, the thought of using broadband for business connectivity gives many network managers the heebie-jeebies, so most SD-WAN vendors promote the concept of a hybrid WAN where MPLS and broadband are both active. The concept is to have mission-critical or real-time apps like VoIP and video use the private IP service, with best effort traffic like many cloud apps going directly to the Internet over broadband -- at least that's the theory.
As compelling as SD-WANs are, there are some questions that deserve exploring further. Here are the top issues that I hear about from customers with respect with making the leap to this new network:
- For global or even national organizations, there is no nation-wide broadband provider. In fact, by my calculations, there are over 800 broadband providers in the U.S. alone. How should large companies choose the best broadband provider(s) to meet their needs?
- Broadband performance can vary quite widely by type of network (DSL, Cable, or Ethernet) or even time of day depending on the congestion level. What should the expectation be with using broadband for business purposes?
- SD-WANs and data center SDNs are often treated as separate silos. Is this the best approach, or should businesses look at a larger SDN strategy that encompasses the data center all the way to the branch edge?
- Making any kind of change to a network can be disruptive, particularly in a world that is becoming more dependent on it. What are the best ways to migrate to an SD-WAN while minimizing risks?
- There are many technologies that fall under the overall umbrella of an SD-WAN, such as WAN optimization, path selection, QoS, Network Functions Virtualization (NFV), and others. Which of these does a business need today?
Answering these questions isn't easy, but I'll be attempting to do so during my panel on the topic at Enterprise Connect, coming up March 27 - 30 in Orlando. I'll be joined on stage by Jason Gadsby, Director of Enterprise Networking at AT&T; Sanjay Srinivasan, VP and Chief Technology Architect Business Engineering at Vonage; Fabrizio Fiori, Global Director, Enterprise Business Solutions, Office of the CTO, Avaya; and Michael Wood, VP of Marketing at VeloCloud Networks.
The panel, "SD-WAN: Why It's the Right Fit for the Cloud Era," will be at 8 a.m. on Tuesday, March 28th; so get up early and come join us to get answers to the above questions and many more.
Learn more about systems management and network design trends and technologies at Enterprise Connect 2017, March 27 to 30, in Orlando, Fla. View the Systems Management & Network Design track, and register now using the code NOJITTER to receive $300 off an Entire Event pass or a free Expo Plus pass.