Traditional approach to WAN is starting to wane as organisations continue on their cloud strategy journeys. Now I don’t believe that home grown data centres (DCs) are going to disappear overnight, but what we have today is a varied mix of multi Cloud, CoLo and On Prem, and will do so for a number of years.
However what this approach to hosting is doing is making the industry rethink how to consume all these hybrid services. This rethink comes at a time when a new set of buzz words is hitting traditional networking. I mean of course SDN (Software Defined Networking) and the more latterly cousin SD-WAN (Software Defined WAN).
SDN is definitely the future, but not what I want to concentrate on today. SD-WAN is more exciting at this juncture, and whilst most people will tell you the main benefit of SD-WAN is to vastly reduce your WAN operating costs (some believe up to 2.5 times cheaper), there is also a great opportunity to implement agile WANs that can consume the new array of service locations like cloud & CoLo and deliver them to your edge network.
Underneath SD-WAN are four key principles:
- Provides Telco Agnostic Approach – in essence it can support multiple transport types (MPLS, Internet, 4G/LTE) – (Known in some quarters as Hybrid-WAN)
- Supports Dynamic Path Selection – so can choose the appropriate physical path, and sending critical traffic over appropriate links.
- Provides Application Optimisation – utilising caching and compression techniques to improve application performance. Think of it as Dynamic QoS.
- A Simple Centralised WAN Management – and more importantly removing Telco lock in for all those complex MPLS changes and QoS tweaks that come at a cost.
There are other facets, but these principles drive a lot of the benefits that organisations are just starting to see from adoption.
There is a lot of technology out there that can deliver SD-WAN (Gartner already identified 23 key vendors for SD-WAN back in May 2017) and the battle for the WAN edge is hotting up. Traditional router hardware will give way to vCPE (Virtual Customer Premise Equipment) along with NFV (Network Function Virtualisation) all sitting on some x86 server tin with a competent hypervisor. The NFVs will sit alongside the vCPE providing what is basically a “branch in a box”. (And yes vCPE is in reality one type of NFV, but we’ll let the tech marketers have their moment). Rapid deployment will be a key aspect of such solutions with a price point that should be competitive when compared with multiple hardware specific solutions. And then there’s the space and power saving aspect, which is not to be sniffed at especially in the cost sensitive retail market.
I believe we are at a key juncture in networking that demands a review of organisation’s roadmaps for their WAN. The gestation period for a large enterprise to consider a change of WAN provider, let alone the technology as well, is long and the actual transformation can be longer. However one of the benefits of SD-WAN is you can “dip your toe” into various solutions whilst maintaining your traditional WAN. Organisations need to start seeing how their teams can evolve into this paradigm shift in WAN management, and along the way understand whether in-house or managed is the way to go.
Whilst traditional managed WAN services will ultimately decrease, I believe we will start to see the likes of traditional VNOs (Virtual Network Operators) start to evolve their offerings by providing a Managed SD-WAN. (Some have been out there for a while just take a look at what Talari Networks are doing stateside).
So ultimately it’s time to stop thinking MPLS = WAN, and to start being driven by software!