IT Brief Australia - Technology news for CIOs & IT decision-makers
Story image
Here’s how broadband CAN be used for critical WAN tasks
Wed, 27th Jul 2016
FYI, this story is more than a year old

Emerging software-defined wide area networking (SD-WAN) technology is challenging the perception held by many that they cannot use Internet connections for meaningful network transactions. SD-WAN has graduated to hot topic status in tech circles, and enterprises are turning to this new approach to resolve their connectivity issues.

SD-WAN is capable of enabling the use of multiple forms of connectivity, it's easy to deploy by virtue of zero-touch provisioning, it centralises management and facilitates significant cost savings.

Consider a connected world in which broadband such as DSL, cable or LTE is so reliable that it is fully incorporated into even the most critical of corporate network connections. Currently it is possible to shift that low cost Internet connection that spends its days as an obscure backup into an integral part of any organisation's WAN. It is also possible for enterprise IT teams to dust off backup-only broadband connections – which they are already paying for, but using only in the direst of circumstances – and make them pull their full weight in a WAN environment.

The solution lies in an SD-WAN. Ordinarily, broadband is regarded as falling far short of the performance, security and reliability required for essential data transfers and business applications, but with the right SD-WAN solution that unpredictable line can be made to perform with excellent quality-of-service levels. It can even be made to deliver increased reliability and robust security.

An SD-WAN abstracts multiple, completely different forms of transport connectivity and combines them into manageable software components. Much as virtualisation software reduces physical servers to functional data instances, SD-WAN tools combine network elements into virtual bundles called business-intent overlays. These virtual overlays take charge of specific applications and apply business logic to them to get the best out of the WAN, allowing admins to cleverly balance cost, quality and availability on an application basis.

For example, for conference calls using VoIP, the most important deliverable is likely to be quality, so an SD-WAN solution should allow organisations to build a business-intent overlay that uses a full-mesh MPLS/LTE deployment. More critical data transmissions require maximum availability. And for minimal-cost services such as Internet connections for corporate guests, another overlay that manages a simple hub-spoke Internet topology will be sufficient.

So to use all their network connections efficiently, enterprises need to select an SD-WAN solution that will truly improve the resilience of their WAN and automatically balance traffic across numerous network paths, all while optimising application performance.

The virtual overlays achieve this by decoupling the logical functions of a network from its underlying physical infrastructure. The value-add for the business is clear: IT admins can concentrate on planning, designing and building, while the SD-WAN heavily mitigates the headache of implementation.

When building overlays, security is baked in at the lowest level. Secure connections, using AES-encrypted IPSec, are built across each physical link before they are bonded together as a single logical connection, leading to enhanced integrity. If a physical link fails, the bonded link, which exists as a software abstraction, remains up and packets are dynamically rerouted to the remaining active links in the bond, all in under one second making it transparent to application users. Because the software behaves in this way, maintenance to any one physical link does not cause downtime, since the logical component will continue to operate normally.

Tunnel and link bonding are not new approaches, but some solutions providers are recognising that if bonding is combined with technologies such as dynamic path-control (DPC), path conditioning and dynamic rate-control, quality-of-service levels can be tweaked on any type of bandwidth, including broadband.

Path conditioning in particular serves as a guardian of transmission quality. It employs two separate techniques (forward error correction and packet order correction) to reduce instances of packet loss and out-of-order packets across a WAN. Forward error correction (FEC) is responsible for the reconstruction of lost packets, which restricts the need for retransmissions and reduces throughput degradation. Packet order correction (POC) can tell if packets arrive at a destination in the wrong order and can automatically reshuffle them as needed – another means of preventing costly retransmissions and throughput degradation. Path conditioning is especially useful in augmenting the quality of weaker links like broadband.

It is when multiple techniques are applied in a seamless and integrated fashion that business-intent overlays can deliver the option of link-bonding policies to businesses. This is where cost savings and quality surges can really be seen by end users. By choosing from high availability, high quality, high throughput and high efficiency, businesses can really put their WAN to work, managing resources in a way that was not previously possible and putting their broadband connection to real, practical use.

Article by Peter Skarlatos, systems engineering manager, Silver Peak Australia - New Zealand