Distributed data architectures require lock-step synchronization and rapid data throughput if they are to have any hope of matching the speed and responsiveness of local infrastructure.
Within the data center, both at the enterprise and in the cloud, software defined networking (SDN) is poised to make it easier to reconfigure and repurpose network architectures on a whim and push bandwidth utilization up to near perfect levels. It only makes sense that to get this kind of functionality across geographically distributed resources, the wide area network will require the same level of abstraction. This is where SD-WAN comes in.
As CTO Advisor’s Keith Townsend notes, SD-WAN is the ideal solution to replace legacy MPLS links that connect today’s virtual islands, and it also provides a range of policy-based routing and other techniques not available on VPNs to optimize the wide area for Big Data and other emerging initiatives. And with an intelligent network controller at the helm, packet analysis, policy decisions and a raft of other functions can take place with little or no human involvement.
A key element in all of this is orchestration, and this is proving fertile ground for a wide range of development. Glue Networks recently showed Ver. 2.0 of its Gluware platform, which offers custom, network-aware design and provisioning aimed at bringing WAN management into the realm of dev/ops rather than lengthy engineering processes. The software includes a new control module that offers network browsing and provisioning in highly scaled environments, plus a lab component that serves as a sandbox for experimentation and innovation. At the same time, the system provides for real-time change visibility, plus centralized orchestration that reduces controller sprawl.
Meanwhile, Silver Peak is out with the Unity EdgeConnect device, available in both hardware and software configurations, both of which provide virtual overlay networks on the WAN within seconds while removing many of the connectivity points that introduce latency between home offices and branch sites. The system also features Dynamic Path Control for policy-based traffic management, plus automatic failover and 256-bit AES encryption.
SD-WAN is also highly adept at determining which networking solution is appropriate for specific traffic based on context, value and other factors, says my colleague at ITBE, Paul Weinschenk. With this ability, the system can select from a range of MPLS, Metro Ethernet and even cellular and broadband links, according to the networking needs of the application as well as the enterprise's budgetary limitations and other requirements. It can even mix and match different network approaches based on geographic and other circumstances. And as mentioned above, it can do this in real time in highly dynamic data environments with broad ranges of latency, load types, QoS requirements and other factors to consider.
For anyone who needs a primer on SD-WAN, VeloCloud recently published “Software Defined WAN for Dummies,” with guidance on basic concepts, deployment strategies, management and operational issues, plus explanations on how the technology improves business agility and data functionality. The book is written for both technical and non-technical readers and provides an easy-to-read style and format. A digital version is available here.
SD-WAN adoption will likely be driven by cost more than anything else. As Big Data and the Internet of Things start to flood the data center and remote sites with increasingly heavy loads, the costs of expanding and maintaining traditional WAN topologies will become increasingly burdensome, if not altogether unbearable.
SD-WAN provides the enterprise what it needs most: a robust, dynamic solution to maximize bandwidth utilization, boost throughput and provide the flexibility to turn the WAN into a geographically distributed LAN.