For WAN managers, network reliability and high QoE for the applications that run over the network are what best define business success. With the advent of SD-WANs, certain vendors and analysts incorrectly say that when combined with their simplified network management technology, the “low road” of the unaided public Internet reliably delivers real-time applications like Unified Communications and mission-critical apps like ERP.
Meanwhile, other vendors claim that the “high road” of retaining MPLS forever is necessary for network reliability and application QoE, falsely contending that a multi-WAN Internet fabric is incapable of delivering QoE because Internet underlay connections themselves cannot offer QoS guarantees.
While the Internet on its own cannot deliver reliability and high QoE, MPLS is neither a standalone “diamond lane” cure nor even a necessary condition for ensuring application sessions do not get broken and perform as users demand.
For these reasons, both premises are wrong. There is, in fact, a “middle road” to take for SD-WANs — one that simultaneously lowers risk while raising the levels of reliability and QoE, as well as other benefits, such as: improved site-to-cloud access, Internet economics, avoidance of vendor lock-in, and, in fact, higher QoE than that possible with the supposed “high road.”
A self-driving, self-healing SD-WAN navigates the best route between these plausible-seeming extremes. But doing so requires sophisticated SD-WAN technology, which most vendors simply cannot offer. Those vendors who tout “application-aware” capabilities are often talking of just identifying applications, leaving it to network administrators to set thresholds and map applications to specific roads.
Network administrators should not be burdened to do what the SD-WAN is supposed to do, which is to direct applications to the optimal paths right now for better reliability and application QoE. Here are four “middle road” capabilities to look for:
This “middle road” approach outlined above offers a near-infinite number of mirrors, onboard sensors, road sensors and destination sensors to monitor what is happening on the road, as well as the ability to route pertinent information to an automated “traffic cop” that directs applications to their optimal paths.