The benefits behind the “software-defined” label are clear: Great advantages are gained by separating the data that people and machines send through a network from the data that controls how it all traverses that network.
The software-defined networking (SDN) concept is a win/win. Networks become more flexible and both capex and opex shrink because less expensive, generic hardware is used. Changing the mission of a network element or a data path no longer involves changing out equipment.
The concept has gained the most traction to date in data centers: concentrated islands in which the distance between networking elements is minimal (though this is changing gradually as data centers become more distributed) and the list of networking protocols used is short.
In short, SDNs are getting their feet wet in data centers because that’s the easiest place to work out the kinks. “The term itself came out about a year ago,” says CloudGenix Co-Founder and CEO Kumar Ramachandran.
Moving SDN Beyond the Data Center
The value of the software-defined concept, however, is just as promising outside the data center. During the past year or so, the term software-defined wide-area network (SD WAN) has become popular. As the name implies, the idea is to export the advantages from the data center to the wide-area network. “If you have only an island of SDN in the data center, but not across the enterprise, you don’t get the [operational] benefits and cost reduction,” Ramachandran says.
It’s not easy, though. Some software-defined equipment is available, but carriers and vendors are just getting serious about developing the broader SD WAN concept. Conversations with experts suggest that there is a consensus emerging on what SD WANs will do, but no consensus on how they will do it.
An SD WAN, experts say, is an approach in which the data plane and the control plane are separated. That, of course is what it has in common with a data center SDN. An SD WAN will treat the wide-area network as a unified fabric, not a series of incremental steps that eventually get packets from one point to another. They will use techniques such as multiprotocol label switching (MPLS) and IP Virtual Private Networks (IP VPNs) and support cable broadband, phone company digital subscriber line (DSL) broadband, 3G, 4G and Wi-Fi connectivity. SD WANs will offer route redundancy, security and the level of quality of service and quality of experience (QoS and QoE) that customers want. Finally, ease of use for the end user is seen as important.
Though all different networking approaches will be managed as one holistic system, SD WANs will retain the ability to segment the network as a way of limiting breaches and taking advantage of a networking technique that offers advantages in a particular situation.
It is unclear if “software-defined” in the name will only be a reference to the overall concept of separating the data from control planes or if it actually will use updated versions of existing SDN protocols.
“There is talk of that, that is something the thought leaders and people driving standard bodies [are aiming for],” said Michael Wood, the vice president of Marketing for VeloCloud. “They are interested in putting together open standards so that they [data center SDNs and SD WANs] can more seamlessly interconnect.”
Much of the mandate that will be given to SD WANs now is carried out by MPLS. This approach routes packets – no matter what protocol they use (hence the term “multiprotocol”) — through the carrier network, assigning them the best path to travel and guaranteeing the quality of service and other attributes promised to the customer.
SD WANs are seen as a way to do this without going through the carrier network, which is an increasingly inefficient approach in a cloud-based and mobile world. Instead, SD WANs will use the Internet.
“I think the real reason it’s so exciting is not so much the software-defined network term, but that it is not built with MPLS,” said David Hughes, the CEO and founder of Silver Peak. “MPLS is not a good match with data going to the cloud.”
It is still early in the SDN game, and the ways in which vendors are approaching the SD WAN sector differ. Of course, the approach that vendors take will, as it always does, depend greatly on what each brings to the table: Is it a startup that can architect its products from the ground up? Is it a vendor with an installed base that must find a way to fit its existing products into the SD WAN landscape?
The bottom line is that the only certain similarity between SD WANs and SDNs is that the control and data planes will be separated. To an extent, the definitions can form over time.
“We are looking at solving a problem CIOs have spoken about for the last six years,” says Ashwath Nagaraj, the founder, CEO and CTO of Aryaka. “Fortune 500 CIOs do not care about terminology. They say, ‘These are challenges I have: The WAN is very complicated, high complex, and we are unable to make transition to SaaS when we move to the cloud.’ Everyone is trying to that.”
Carl Weinschenk covers telecom for IT Business Edge. He writes about wireless technology, disaster recovery/business continuity, cellular services, the Internet of Things, machine-to-machine communications and other emerging technologies and platforms. He also covers net neutrality and related regulatory issues. Weinschenk has written about the phone companies, cable operators and related companies for decades and is senior editor of Broadband Technology Report. He can be reached at [email protected] and via twitter at @DailyMusicBrk.