There’s been a ton of recent vendor activity in and around the enterprise WAN. For starters, there are several start-ups emerging like Glue Networks, Viptela, CloudGenix and Velocloud. Further, Cisco announced a bunch of new hardware/software in conjunction with their IWAN architecture at Interop in the Fall, including a much-needed revamp of their path selection software (PfR). Silver Peak launched their Unity software architecture late last summer, which is essentially an IPSec-based SDN overlay that runs over the Internet and uses “cloud intelligence” to pump information back into the forwarding decision. In addition, Nuage Networks (a spin-in from Alcatel-Lucent) recently extended their data center overlay solution to the WAN as well. And these are just a few…
What all these technologies have in common is that they are in support of the “Hybrid WAN” – utilization of multiple access technologies (typically MPLS and Internet) to achieve optimal cost and performance for the enterprise. The value proposition associated with Hybrid WAN and the technology-solutions mentioned above include:
- Reduced WAN costs via using lower-cost Internet/broadband (to supplement, compliment, and/or replace traditional MPLS).
- Simplified and improved management orchestration of WAN traffic/devices, including path selection capabilities (i.e., less MacGuyery traffic engineering).
- Improved and unified visibility and monitoring traffic.
- Improved security, including improved scale for VPN-based solutions.
This very well could be the elusive “killer” SDN use-case, as many clients I speak to cite major challenges with their WAN (Cost, Performance, Availability for starters). Along these lines, we are planning to publish more in-depth research on this topic in 2015, but in the interim you can check out this highly relevant research on Hybrid WANs, authored by my colleague, Bjarne Munch:
Hybrid Will Be the New Normal for Next Generation Enterprise WAN | Bjarne Munch
Summary: Public cloud services and mobility are driving business needs where the Internet and MPLS play an equally important role for enterprise connectivity. Network planners must establish a unified WAN with strong integration between these two networks to avoid application performance problems.
The Gartner Blog Network provides an opportunity for Gartner analysts to test ideas and move research forward. Because the content posted by Gartner analysts on this site does not undergo our standard editorial review, all comments or opinions expressed hereunder are those of the individual contributors and do not represent the views of Gartner, Inc. or its management.
This article lacks a fundamental understanding of how most enterprises consume MPLS and WAN technologies in the current ERA. Enterprises typically do not operate MPLS themselves and the concept of traffic engineering is almost exclusively the realm of Service Providers. Path selection in a hybrid MPLS / Internet WAN does not require SDN. It is a very simple case available today of using routing protocols at the CE end of the service tails. Differential and policy based forwarding are also options at this point, the only way you could link SDN to these functions would be the automation of the push of these changes to the devices, and this is not SDN, it’s just config automation.
Martin, thanks for reading and commenting. I appreciate your comments and respect your perspective on things. It is always good to get feedback like this, as it provides an opportunity to further clarify. I’ll add some points to provide some additional context.
Point of clarity: I was not referring to traffic engineering within the carrier’s MPLS network. I was referring to traffic engineering from the perspective of inbound/outbound enterprise traffic @ the WAN edge. This would be traffic across multiple transport (MPLS, 4G, Internet, etc) to apps on the public Inet and within corporate data centers. I agree with you that traffic engineering within an MPLS carrier is not an major enterprise concern.
With regards to Administration/Management of enterprise WAN equipment, it is dramatically different depending on Geography. In North America, most large enterprise prefer to manage their own CPE (this is based on thousands of client interactions). In LATAM and Europe, the carrier manages the majority of CPE. APAC is not homogenous, it varies. So the net net is that in certain GEOs, these SDWAN approaches provide direct benefits to a large number of enterprises.
That said, from the carrier perspective they are also interested in these technologies. Many of the carriers are conducting SDN/NFV trials within the environments. Further, several large carriers are piloting and in limited production with some of these newer SDWAN players.
On a related note, I agree with you that SDN is not required. Orchestration and automation can provide many of the benefits outlined above. There is a blurring of terminology and we are in the process of writing research on SDWAN to help clarify. But a perfect example would be someone like Glue Networks; they provide orchestration software that addresses these challenges but it does not meet our definition of SDN.
This article was very useful for my Term paper which is based on the WAN topic. I would like more details, however it was on point. Thank you Lerner.
thanks, happy to help
Thanks for the very informative info. i have a networking project and i’m not sure what direction to look. i’ve been looking online and i’m trying to gather an estimate for a LAN/WAN network for 50 employees, 5 sites with 5 remote consultants. thanks for the help.just basic cost.