Sponsored By
The cloud -- the union of cloud computing, the Internet and software-as-a-service (SaaS) -- is rapidly changing enterprise IT. The traditional enterprise WAN solution of the last decade -- MPLS, often augmented by WAN Optimization technology -- no longer cuts it. A new technology, Software Defined WANs (SD-WANs), offers enterprises tremendous opportunity for cost savings, flexibility, bandwidth, manageability and superior cloud access. But SD-WAN implementations, while sharing a few common traits and tending to sound alike, in fact vary significantly.
MPLS is still a $15 billion plus worldwide market -- despite the enormous price/bit gap versus Internet connectivity -- because MPLS offers "failsafe" WANs. Failsafe here means the union of high availability and reliable, predictable network performance and application Quality of Experience (QoE). Yet it turns out that only a few companies have actually developed failsafe SD-WAN technology.
MPLS doesn’t offer connectivity to public cloud computing and SaaS. MPLS locks you into a single service provider, and is incredibly expensive, as it costs literally 50 to 150 times more per Mbps for smaller sites, and can be 10 to 50 times more expensive per Mbps at data centers. Meanwhile, traditional WAN Optimization technology offers relatively little benefit over high bandwidth but unpredictable Internet WAN connections, especially as less and less traffic is Microsoft CIFS file service.
SD-WAN implementations from most vendors do share some things in common: support for multiple WAN links creating a flexible virtual WAN overlay fabric, not tied to any single service provider, that can leverage lower-cost Internet links; centralized controllers with centralized management, reporting and policy control; a choice of physical and/or virtual appliances.
But vendor SD-WAN implementations differ widely. Some focus on the needs of telecom service providers. Some focus primarily on "router replacement" (the cost of owning/maintaining the routers), rather than the issues surrounding the unpredictability of Internet connectivity, cloud access and the costs of WAN service. Some focus on offering a fully managed service over the public Internet. While these last can be fine for some smaller enterprises who have never had MPLS, nor likely any kind of centralized security architecture, they can be of limited value for larger enterprises with greater control and security concerns, or who do not wish to make an abrupt wholesale cutover of their WAN infrastructure. Few SD-WAN implementations have focused on failsafe WANs.
Why failsafe SD-WANs? Failsafe WANs means both high network availability and reliable, predictable application QoE. The tantalizing benefits the cloud and SD-WANs might promise notwithstanding, enterprise IT folks disdain "two steps forward, one step back" solutions for good reason. Even the most aggressive, leading-edge enterprises want failsafe WANs if they are going to augment or replace their time-tested MPLS-based solution, given the high cost of downtime, limited IT resources available at most remote locations, and limited enterprise visibility into the WAN.
What are the key SD-WAN technologies that enable the delivery of failsafe WANs? The linchpin is continuous unidirectional measurement of packet loss, latency, jitter and bandwidth utilization for all paths between any pair of locations. Continuous measurement of each packet on the WAN, as opposed to measurement via occasional round-trip test packets; unidirectional, since just as on roads and highways, traffic jams are usually in one direction but not the other; across all paths, since given m Internet links at one location and n at the other, there are m x n possible network paths that can be used across the network-of-networks that is the Internet, substantially increasing reliability and QoE.
This precise, real-time monitoring enables unprecedented WAN visibility and sub-second response to adapt to network issues, both hard link/device failures as well as "soft outages" from network congestion-based packet loss and jitter. It is this sub-second response that is the critical component to delivering a truly failsafe SD-WAN. Other important failsafe SD-WAN capabilities include: the ability to use all bandwidth across multiple links, even for a single high-bandwidth flow when doing link aggregation; optional packet replication for real-time traffic, providing platinum quality real-time support; highly efficient bandwidth utilization -- delivering 90% to 97% sustained utilization on shared inbound links without negatively impacting QoE (all reputable basic QoS solutions have been able to do this for outbound links for years; doing this for inbound traffic is much more difficult); and high QoE cloud access, especially in combination with carrier neutral colocation facilities close to the core of the Internet. Finally, pay attention to the ability of an offering to scale to hundreds of locations and thousands of WAN links (not trivial to deliver when doing continuous measurement on all network paths and offering sub-second response). What can enterprises expect from a properly designed failsafe SD-WAN? Benefits include, but are not limited to:
SD-WANs offer enterprises enormous potential and benefits -- if delivered in a failsafe manner. A failsafe SD-WAN solution offers enterprise WAN managers a safe, evolutionary, cost-effective way to deliver QoE while meeting application demands in the age of the cloud. Few vendor SD-WAN solutions actually deliver failsafe enterprise WANs. Make sure to source your SD-WAN from a vendor that does.
