SD-WAN – has it lived up to the promise?

Software Defined Wide Area Networking, or SD-WAN, first started to gain real traction around 2017 following its entry in to the Gartner Hype Cycle for Networking and Communications in 2016. The communications industry heralded it as the next generation of networking; the biggest paradigm shift since the late 1990’s when Multiprotocol Label Switching (MPLS) was introduced.

Some pretty wild claims were made around the benefits that it would bring to customers, both from vendors and service providers alike.

Now that SD-WAN is becoming mainstream, I thought it was about time we looked back at some of those original claims and see whether it has justified the hype.

Claim 1 – the death of MPLS

Multiprotocol Label Switching (MPLS) is the technology that most network service providers use to offer multi-tenant access to their private networks. It means that we, and businesses like us, can operate a global private network and share it between thousands of customers without putting their traffic at risk from one another.

SD-WAN was supposed to kickstart a mass migration from MPLS services to the Internet. Internet services are typically cheaper than private MPLS because the Internet is a ‘best efforts’ network. In other words, you’re unlikely to get a guarantee to cover how well your traffic crosses the Internet, but you would for most MPLS connections.

SD-WAN has built in capabilities that get around some of these limitations and this was seen as the catalyst for an Internet-first approach.

The reality has been subtly different. We have seen a hybrid network rise to the top of customer’s wish lists, where we use a combination of MPLS and Internet connections to forward traffic in the way that best suits each application.

Some services are naturally bound for the Internet, like Microsoft’s Office 365 and Google’s G-Suite. Other applications need the speed and reliability of a private MPLS circuit.

SD-WAN is aware of which application needs which type of connection and distributes the traffic accordingly. This hybrid approach tells me that MPLS is far from dead and is looking like it will be around for a good few years to come.

Claim rating: 3/10 – the Internet has taken share from MPLS, but private networking still offers some compelling benefits for those that have the need and means for it.

Claim 2 – customers will want to take control of their own network

SD-WAN is a simpler proposition to manage than its predecessors. Almost all vendors offer centrally hosted, portal-driven management where administrators can point and click their way to network management nirvana. This is a huge step forwards from the command lines of the previous generations of infrastructure.

With this decrease in complexity, many people believed that customers would be flocking to take control of their network estate. This would effectively allow them to commoditise their network circuits, putting them in a stronger bargaining position when negotiating with their network service provider.

There was an initial surge in customers taking this route, but it hasn’t lasted. There are a few reasons why but a lot of it comes down to risk and reward.

Network companies like ourselves have a long history of managing mission critical networks and dealing with the unique challenges that brings. It is more than a collection of circuits and, because of this, many customers have switched back to a managed service.

The exception to the managed services model is in North America where companies regularly look after their own networks. On average though, we’re now seeing about 80% of customers preferring managed or co-managed SD-WAN services versus doing it themselves.

Claim rating: 4/10 – some customers have taken the leap, but not in the numbers that were predicted.

Claim 3 – big reductions in operating costs

There were expectations that organisations could slash their network budgets by over 60% with a move to SD-WAN.

To make this sort of saving would require a significant hit in either the performance of the network or the assurances that you got from your service provider. Working with some analysts, we have estimated that average savings are around 30%. This is based on some sites moving to more cost-effective access methods such as the Internet.

The other area for cost savings is around reduction in managed services charges. This is dependent on the level of involvement each organisation takes in running the network so is correspondingly variable.

Claim rating: 6/10 – SD-WAN has delivered operational savings, but the value is based on each organisation’s circumstances.

Claim 4 – much greater agility

I’ve mentioned that SD-WAN is managed from a central portal and this is the key to the agility improvements.

Administrators can create a set of configuration changes once and push them out to lots of devices at the same time. This is particularly beneficial to larger organisations with hundreds or thousands of sites to manage.

The other side of the agility improvement is linked to the approach an organisation takes to management. If they take on some or all of the management responsibility, then they can push the changes out in near real-time without having to raise a request on their service provider.

This is a real step change in reaction time and allows networks to be much more responsive to security threats and changes in business requirements.

Claim rating: 9/10 – there’s no doubting the impact SD-WAN has had on modernising the network and bringing the responsiveness in to the 21^st century.

Claim 5 – network functions will rule the world

Network Function Virtualisation (NFV) describes the way we can run applications within an SD-WAN.

It was originally built on the same virtualisation approach that allowed us to run multiple applications on a single server in a data centre. Effectively, turning the router we use to connect a customer to the network into a multi-purpose box that runs a virtual router amongst other functions.

The hope was that organisations could start to:

• Run virtual firewalls to protect their branch locations,
• Implement traffic optimisation to improve network performance, and even
• Run some local file storage on site to allow for the retirement of on-site computers

To reflect this change in status, we renamed the router to a universal Customer Premises Equipment, or uCPE for short.

However, when the industry started to get into the nuts and bolts of hosting more functions on the uCPE, each box needed more horsepower, space and cooling on customer sites – both of which drove the price up.

So, NFV’s running on a customer site wasn’t the panacea that the industry had hoped for. The good news now is that there is a better place to run network functions: the network itself.

We can use technologies like distributed Multi-access Edge Computing (MEC) which delivers public cloud agility and functionality, making it elastically scalable. It can also be designed from the ground up to support functions for multiple organisations.

Claim rating: 7/10 - based on the promise and performance of MEC rather than relying on the uCPE.

So, what’s the verdict?

Well, it’s a bit of a mixed bag. Some, such as the death of MPLS, were exaggerated; some, such as the rise of NFV, were a bit premature, and others, such as the improvements in agility, were right on the money.

All in all, when you compare the benefits of SD-WAN against a more traditional network approach, there’s no doubt in my mind that it’s better for the vast majority of customer requirements.

It brings a step change in visibility and agility, coupled with the flexibility to use a mix of access types to suit any location. It’s also an important step on the journey to future architectures such as the Secure Access Service Edge and SD-Branch. Why don’t you give it a try?

Learn more about what SD-WAN can do for your business.