Now that the enterprise has gained a little experience pushing select applications and services out to the cloud, it’s time to start thinking about integrating this new infrastructure into the overall data environment.
This usually calls for the data center to become more cloud-like, and most organizations are well on the way to making this happen through the advanced pooling and management capabilities available on virtualized infrastructure. But that still leaves the link between local and distributed clouds, which, if not designed properly, can act as a bottleneck between virtualized networks both at home and at remote facilities.
This is why many organizations are warming up to SD-WAN technology. Using abstract software overlays similar to standard SDN, the wide area network can be made as functional and fungible as the local network, ushering in the possibility of a truly distributed yet fully integrated software defined data center (SDDC).
IDC pegs the annual growth rate for SD-WAN technology at 90 percent for the rest of the decade, topping out at $6 billion by 2020. Specific product categories run the gamut from application-based policy controllers and software overlay platforms to the advanced analytics and visibility tools to enable highly dynamic and automated network performance. Beyond that, organizations will also likely invest in intelligent forwarding systems to navigate the multiple pathways that will arise across disparate WANs. If all goes well, the result will be a more cost-effective means of delivering business applications, as well as highly optimized performance of SaaS and other cloud-based services.
Already, developers of the various SD-WAN pieces are coming together to foster more integrated solutions. Viptela and SevOne recently joined their respective SD-WAN and Digital Infrastructure Management Platform offerings to enable high visibility across virtualized WAN architectures. Viptela users will now be able to use the same SevOne dashboard that oversees local infrastructure to gauge the health of the wide area network. This is a crucial step for many organizations considering that a software-defined WAN will not be the rigid construct of today but will take on many of the LAN-like functions that exist within the data center. Managing the entire architecture as a single entity will allow application and services environments to transcend the local environment to sit more comfortably on a distributed cloud.
Much of the SD-WAN will function as an intelligent, self-managing environment, particularly as Big Data and IoT operations start to push higher volumes to and from the network center and the edge. As Glue Networks’ Stefan Dietrich puts it, “Manual processes are simply too expensive and time consuming. IT governance suffers and the organization struggles to maintain consistent service levels across all lines of business.” A logical IP network is helpful, but to fully realize the technology’s potential, you’ll need a network-aware orchestration stack that enables best-practice architecture provisioning, high-speed change management, configuration monitoring and advanced trouble-shooting. And it goes without saying that as technology is updated, network technicians should update their skillsets, too.
Have a good idea of what you want the SD-WAN to accomplish before you get carried away by the technology. Talari’s Kevin Gavin identifies five key benefits that the SD-WAN should provide if it is to play a crucial role in the distributed data ecosystem. These include maximizing network choices and working across multiple WAN architectures to provide the flexibility needed to support advanced user-driven and M2M processes. It should also support a wide range of application requirements so business productivity is no longer hampered by bandwidth or connectivity issues. And, of course, it should lower costs, which, on the WAN, can scale up rather quickly.
With data loads being what they are these days, and about to get worse, many CIOs are starting to realize that one data center cannot handle it all. A scalable, distributed architecture is the only viable avenue at this point, but not if it is hampered by an expensive, inflexible wide area network. Layering static hardware with innovative software constructs is the only logical answer to what the enterprise truly needs: a fully distributed data ecosystem that can be crafted on-the-fly to take on any load that comes along, no matter how larger or how complex.