One of the biggest developments in enterprise networking at the moment is the emergence of the software-defined wide area network (SD-WAN).
This next generation wide area network (WAN) technology is being touted as a means of moving away from proprietary or specialised WAN technologies to something more open, flexible and cloud-based.
Frost & Sullivan recently revealed that 94% of businesses have deployed, are deploying or will deploy an SD-WAN in the next two years. So, it looks like SD-WAN is gaining traction, but what is SD-WAN and how will it change the humble enterprise WAN?
Simply put, SD-WAN streamlines enterprise connectivity between remote locations and branch offices, using software to create encrypted network tunnels between endpoint devices to produce a private WAN over simple circuits.
It uses broadband internet, 4G Long Term Evolution (LTE), or multiprotocol label switching(MPLS) access and centralises control in the cloud.
The upshot of all of this is the enterprises should, in theory, gain greater flexibility and performance in their corporate network as well as better economics.
Benefits of SD-WAN
How is SD-WAN going to make IT operations more effective? Gina Nomellini, chief marketing officer at network provider GTT, says that as IT and applications today are often more geographically diverse than their user base, SD-WAN is helping make the network simpler to configure and control centrally.
“SD-WAN allows you to consolidate your network assets – old and new – and extend control all the way to the edge for more efficient branch networking and secure connectivity to cloud applications,” she says.
She adds that using split-path routing and dynamic-path control, SD-WAN lets enterprises route their internet traffic locally, without having to re-route through the WAN and squeeze through the datacentre-hosted security appliance. It improves cost efficiencies and enhances networking flexibility, without sacrificing one of the most important business drivers, performance.
Richard Kitney, hybrid connectivity specialist at Orange Business Services, says that a compelling initial benefit will be in switching on application visibility before any business goes headlong into delivering services to the cloud or re-architecting to a hybrid network topology.
“That level of visibility gives a lot of usage detail about what applications are running on the network, how much bandwidth is being consumed and where any problems lie. This in turn enables the business to have an informed discussion about current applications and how they can be improved,” he says.
Fitting in with existing technology
If a business already has inter-site connectivity, either through site-to-site VPN, or MPLS, then it will probably already have technologies in place on the network to support multi-site implementation of SD-WAN, according to network infrastructure design firm LAN3’s SD-WAN team manager, Roger Collins.
“SD-WAN solutions, that include a dedicated backbone, will provide the benefits of MPLS at significantly reduced cost while providing reduced complexity and ease of administration (when compared with site-to-site VPN solutions). The technical prerequisites will be similar to these existing technologies and will include ensuring non-overlapping subnets between locations,” he says.
Donna Johnson, vice-president of product and solution marketing at Cradlepoint, says that the main thing an organisation should think about before implementing SD-WAN is to really understand their use case – what are the primary problems they’re trying to solve or what do they think they’ll accomplish with SD-WAN?
“Because the technology itself is so broad, it’s important to go into a project with a definite idea of their end goal. Without that, it’s easy to get lost in the very many different options and technologies that are included in SD-WAN,” she says.
Finding the right SD-WAN
Marc Sollars, CTO at integrator Teneo, emphasises the need for education on SD-WAN capabilities and risks.
“Vendors need to help prospects identify potential SD-WAN risks and what outcomes they need because each vendor goes about this discussion differently,” he says. “It’s possible to have some zero-touch SD-WAN deployments, because of a rapid go-to-market plan or a tough deadline, but I know of roll-outs where networks fell over because the risks were still not fully assessed.”
He adds that the corporate decision-making unit is broad and advocates a workshop approach, ensuring that all stakeholders – CIO, CISO, etc. – are on the same page, in terms of knowledge and business outcomes.
“I know of a transportation company that thought they had every risk covered, but we showed its team additional capabilities that needed further discussion. I know quite a few SD-WAN experts that are saying prospects have had some Kool-Aid from vendors but not the truth,” says Sollars.
Kitney says that enterprises should look out for end-to-end technology. “For example, for quality of service, you really want that to be delivered across the board”.
From branch site, to HQ, to cloud, enterprises should ensure the quality of service can be maintained across the infrastructure.
“It makes things problematic when you’re crossing boundaries from SD-WAN to mixed connectivity environments to cloud and have to translate everything and make it work in an ad-hoc fashion. There needs to be a ubiquitous infrastructure end-to-end,” he adds.
The options for SD-WAN
There are a few approaches on how to adopt SD-WAN technologies. Whilst all SD-WAN vendors will help reduce the costs associated with MPLS, most will achieve this by reducing the volume of traffic which is sent down the MPLS circuit, according to Collins.
“Most appliance-based SD-WAN solutions use the ‘public internet’, where latency is unpredictable, so MPLS is still a necessity for latency-sensitive applications,” he says.
Collins adds that this often requires the maintenance of existing routing devices and an additional ‘overlay’ SD-WAN router/firewall. “This means that budget needs to be sought to purchase these additional devices and resources required to manage the devices,” he says.
Collins says that for companies with several office locations dotted around the globe, latency is a significant challenge and MPLS has, up to now, been the only answer (but at a price!).
“However, there are now viable alternatives, where the savings can be substantial, and little compromise to latency. These approaches hinge on the provision of an affordable, SLA-backed global backbone, where latency is minimised,” he adds.
Sollars says that when it comes deploying SD-WAN, phasing of migrations is essential.
“Many companies want better performance from existing applications in a particular region or division, or they want greater reliability from an enterprise-wide application – such as video, collaboration tools, Office 365, and so on,” he says.
“We have a customer considering SD-WAN for 1000+ locations and we’re carrying out phase one deployment to a few sites. Global-level deployment only will follow when the customer’s IT team is comfortable with application performance.”
He adds that as an alternative, companies that target particular outcomes from SD-WAN, mitigate risks and keep implementation simple, are more likely to be successful. “Some firms only do cover one area of the business to see a clear benefit. One firm switched off its MPLS circuits and moved to public internet circuits to save costs and it’s on target to save $750K a year.”
Read more about SD-WAN technology
- Software-defined WAN technology is available in a variety of business models, including SD-WAN as a service and managed SD-WAN services. But how do these two models compare?
- Companies are approaching network infrastructure upgrades with SD-WAN deployment, but a new survey shows they’re weighing issues like service chaining and integrated functionality.
- IT teams thinking about software-defined WAN technology should assess a few SD-WAN security concerns before deployment to make sure it’s worth the investment.
When deploying SD-WAN the network underlay is a key part of the SD-WAN architecture. It’s complex and often underestimated, according to Mark Weait, VP and head of Europe at Tata Communications.
He says that networks are far more complex than they were a decade ago, with infinitely more applications to consider: some on-premise, some in the cloud, some transitioning between, and many of these are mission-critical.
“These applications need to be secure and free of congestion to avoid QoS issues arising. Businesses starting to implement SD-WAN need clear visibility of all applications, as well as a clear understanding of what should go in which queue and, when congestion occurs, what should take priority,” he says.
He adds that more importantly, where does the low-priority traffic go during busy periods? “When it comes to optimising voice for SD-WAN, architectural underlay is vital for the overlay to work efficiently. This is especially important for businesses looking to reap the benefits of unified communications (UC) and cloud-enabled services such as SIP trunking.”
There will be significant developments and challenges for next generation SD-WAN deployments over the next 12 to 18 months.
According to Manish Aggarwal, the AVP of technology at design and engineering firm Aricent, there are interoperability challenges that need to be overcome.
“The industry requires a set of agreed, uniform interfaces so multiple-vendor components in overall SD-WAN ecosystems can interoperate via standard protocols and APIs,” he says, adding that the MEF industry association has started working on this.
Aggarwal also expects the development of intent-based SD-WANs as this will allow rapid and automatic provisioning of required network elements. “It will use capabilities like AI (Artificial Intelligence), ML (Machine Learning) & Cognitive Analytics to learn the characteristics and performance of network flows/applications and configure the best possible automated paths and policies.”