At midnight on the 31st December 1999 the 20th century came to a close, and amid all the
fireworks and champagne, we said goodbye to our past. A century that began with the
Victorians and silent films, ended with Britpop and the World Wide Web. The Internet was
still new, smartphones and tablets unheard of, and Artificial Intelligence was a science fantasy.
As global networks lit up with pictures and messages from new year celebrations in Hong
Kong, Sydney, New York and London, and the feared Y2K millennium bug proved more
myth than menace, the types of networks used to transmit data long distances were
predominantly Frame Relay, ATM, and the newly emerging Label Switch technologies.
All this will very soon be 20 years ago, and while massive leaps in technology have driven
global advances almost everywhere else, wide area networks haven't changed greatly since
MPLS first came into use at the turn of the century.
So after two decades of MPLS dominance is it time for a rethink? There are often long
lead times on the provisioning of these circuits, an inflexible capability to change network
topologies without major cost, redesign and re-engineering, and little ability to dynamically
adapt to the changing demands of a very much more complex 21st century world. A network
generally has to be sourced from a single carrier who then has a significant lock-in for many
years on network supply and management at a predetermined cost.
The emergence of SD-WAN has challenged this proven model, and is both a disruptor and
facilitator; by cutting costs and making the network a flexible enabler for cloud derived 'as a
Service' functions it cuts across the old order, and the separate management layer can allow a
vendor neutral service partner to source and run best-cost services from multiple telcos and
create a performant network in a very bespoke way.
IT functions today are far from static and predictable, and place massive demands on
networks by changing dynamically every single moment, so the modern WAN needs to
facilitate very bursty on-demand traffic, and have very high performance and resilience.
As M2M, IoT, and AI become a background to all aspects of our lives, traffic levels will
grow exponentially, and add ever increasing requirements for networks to accommodate.
Many companies now use elastic cloud compute such as AWS in lieu of traditional
datacentre hosting for the build of IT systems, so that fits SD-WAN very well too. Cloud
hosted services such as Office 365, Teams, Slack, Skype for Business, Salesforce CRM, and all
manner of SaaS offerings, are also prime drivers for the adoption of SD-WAN, as instead of
all cloud traffic routing via a corporate data centre it can use local breakout with some very
clever use of route optimisation.
SD-WAN can supplement existing legacy technologies, so it can be seen as an evolution
rather than a major change. The use of high cost fixed line circuits alone is no longer the
only option, and their displacement by lower cost direct internet access circuits and
broadband services, in tandem with MPLS, can be used to create hybrid flexible networks
These new hybrid networks can be a mixture of MPLS, DSL, LTE and even cable if
required, and may vary by location, network availability, carrier topologies and traffic types,
often being coupled with WAN acceleration for an added performance boost. Real time and
synchronous traffic such as video, voice and collaboration, can be managed with application
The networks that served us well at the end of the 20th century are unlikely to be well suited
to the world of e-commerce and rapid change of nearly two decades later. The adaptability
and resilience of lower cost and more manageable SD-WAN networks are available to us now,
and will create a springboard for businesses right into the next decade of the 21st century.
Few of us still use our PDAs, VHS players, and 2G mobiles anymore; we have moved on
since the year 2000, and so surely must our networks in this new era.
Author: Gary Dudbridge
Copyright Gary Dudbridge 2019