Build a reservoir of bandwidth liquidity with SD-WAN
Communications service providers (CSPs) are now looking to help large enterprises and public-sector organisations transform the edge of their networks, which now require seamless cloud connectivity and app-performance integration, writes Atchison Frazer, the worldwide head of marketing at Talari. One area of innovation lies in the implementation of NFV strategies or network functions virtualisation (NFV). For example, if a carrier supplies a circuit to an enterprise client – such as multi protocol label switching (MPLS) – they would also like to offer the optimisation of that circuit as a subscription service. Then, the CSP can work with best-of-breed WAN-edge network innovators to create VNFs or virtual network functions of typical edge-network services, such as software defined-wide area network (SD-WAN) or WAN-optimisation. These virtual network functions (VNFs) can be service-chained together to be exposed to end users as one unified service.
One risk to the vertically integrated strategy of MPLS-SD-WAN as-a-service delivered by a telco is establishing a single point of failure – something to avoid in large-scale distributed networks. Primary market drivers to SD-WAN adoption include the need to establish WAN redundancy, ensure that multiple pathways are optimised and to implement platinum QoS standards with multi-level failover continuity – commonly deployed to ensure continuity of communications when a MPLS circuit degrades or drops.
Given that some of the key challenges for network managers considering SD-WAN are increasingly complex, many face a common dilemma. Do they continue to sweat the assets by layering over obsolete legacy equipment and costly over-provisioning, or initiate a complete makeover and digital transformation of the entire WAN? SD-WAN is a flexible component for either strategy: augment MPLS to extend its total cost of ownership (TCO) value, or consolidate multiple branch office devices with one SD-WAN solution. Either scenario acts to create a reservoir of bandwidth liquidity.
One of the other pitfalls to look out for in terms of single-source telco provisioning is so-called vendor lock-in. To move away from legacy MPLS, some SD-WAN vendors require specific carriers, rather than being carrier and link-type agnostic. Some vendors also only offer traditional WAN-optimisation but no real software-defined, controller-based platforms. Unless you can measure performance and routes dynamically at the packet-by-packet level in millisecond real time as apposed to session-steering of bulk packets with round-trip measurement only, then attempting to run SD-WAN for latency-sensitive apps such as VOIP or video-conferencing will be severely challenged.
When making the switch from legacy to SD-WAN, it’s also advisable for network administrators to look for a solution with enough granular edge visibility so that SLA attainment can be accounted for in terms of QoS standards for business-critical app delivery, which includes loss, latency, jitter and bandwidth as well as WAN-link redundancies and disaster recovery.
Administrators require flexible solutions to solve issues of bandwidth allocation, network uptime and failsafe reliability. They need solutions that deliver results, when it comes to eliminating complexity and driving down operational costs associated with multiple devices and achieving carrier-class connectivity from commodity sources.
Digital transformation is accelerating business processes at the edge and creating new opportunities to leverage data in fresh and innovative ways. However, to fully take advantage of digital technologies and their impact on strategic business initiatives, enterprise WANs must remain agile, reliable and cost-effective to successfully support this transformation and not be vulnerable to one-link failure disruption.