Multiprotocol Label Switching (MPLS) turned 21 this year and the ‘grown up’ routing technique is still the preferred WAN infrastructure for many larger enterprises. However, the standard has not evolved since 2011 and pricing, although down somewhat, is still an order of magnitude greater than DSL business internet based alternatives. With bandwidth and application performance demands rising, many organisations want a viable alternative, but moving from MPLS to more cost-effective options – with SD-WAN as the enabler – is sometimes wrongly perceived as high risk.
Emerging alternatives
MPLS has gained its dominant role through its ability to guarantee a high level of predictability in terms of networking connectivity using built-in quality of service for what is essentially a dedicated circuit. However, business-grade broadband is catching up and newer products such as Ethernet leased lines, business cable and SDSL FTTC that are available in the UK through the likes of BT, Zen and Virgin Media can offer many of the QoS and upstream/downstream benefits to varying degrees.
Larger enterprises with distributed sites tend to need more granular capabilities and outside of just lowering cost, SD-WAN allows centralised management across branch networks and deeper visibility into the network. The result is often more responsive applications, while a beneficial by-product is enhanced resiliency across the WAN architecture gained through disparate connectivity paths.
In simple technical terms, SD-WAN intelligently and dynamically works out how to best route application traffic based on the business requirement and available connectivity. So, for example, it can prioritise time-sensitive transactional data over ongoing backup data – or possibly use an underutilised DSL link rather than a saturated MPLS depending on factors like criticality, latency or available bandwidth.
SD-WAN as a concept has been successfully used for about a decade, but the main sticking point for transitioning for many network admins is around the complexity of implementation. The fear is that WAN traffic may be impacted, business may be disrupted, or that physically provisioning new equipment and making extensive configuration changes is a resource-intensive project that thinly stretched IT teams would rather put off – until never.
Not created equally
This perception has been borne out by the different approaches of key SD-Vendors that range from a full-scale rip and replace methodology to the other end of the spectrum offered by newer Transparent Hybrid WAN (THWAN) technologies. At the rip and replace end favoured by the likes of Cisco, the suggestion is to gain full SD-WAN capability which will require the swapping out of key routers and switching with the corresponding wholesale network reconfiguration. For the likes of Infovista’s Ipanema SD-WAN, THWAN uses an overlay technology that sits alongside the existing network and MPLS circuits to offer complete SD-WAN functionalities while eliminating the need for hardware changes, and dramatically reducing network configuration changes.
In terms of practical advice on deploying SD-WAN, organisations need to consider 5 key points. The first is to catalogue any issues around application performance, scale or resiliency that are impacting the organisation. Upgrading existing MPLS circuits for more bandwidth may not always solve the problem if the issue if centred on prioritisation of traffic rather than bandwidth utilisation.
The second factor is to examine what other potential WAN network technologies are available within the geography, and the timescales involved in enablement. Setting up a new business broadband link in major cities is relatively fast, but the current COVID-19 health crisis has led to extensive delays.
