Why WAN optimisation should be a no brainer for SD-WAN

As someone who has been following the evolution of enterprise WAN architectures over past decades, it has been fascinating, especially for the number of new technologies that have been deployed in isolation.

For example, WAN optimisation and SD-WANs are often discussed as separate solutions. From my perspective, it's hard to fathom why a business would deploy an SD-WAN and not implement WAN optimisation as part of it.  If tech professionals are going to go through the work of modernising their WAN architecture, then why wouldn't they integrate optimisation technologies into their deployment right from the start?

In discussions with network managers who have gone down the SD-WAN route sans WAN optimisation, they say something to me like: 'that's for application performance, we are building an SD-WAN to save money'.

After grabbing my head in disbelief, I probe a bit deeper and ask something like, 'shouldn't everything in IT in some way favourably improve application performance?' As a general rule of thumb, I believe that for new technology projects the solution to a problem should never be more complicated than that original problem.

Performance

Let's play out the case where an organisation decides to build an SD-WAN without implementing WAN optimisation. Will the business leader really be happy if it impairs application performance? Any cost savings realised are immediately negated the first time a critical order times out or an important customer call drops. So IT leaders must take into account how users are impacted whenever they deploy new technologies.

For example, several years ago one of my clients migrated a number of critical applications to virtualisation platforms. The CIO sent a message out to the company to communicate how this new project would save the organisation millions of dollars, improve utilisation and mark the beginning of a larger IT modernisation initiative.

In the morning, after the systems running on the virtual machines had gone live, he received call from one of the company's business leaders who had literally timed the performance of transactions with a stopwatch prior to the migration to VMs, and again afterwards. He insisted that the new system was slower and demanded the applications be moved back to dedicated servers.

I really didn't believe the virtual machines were noticeably slower, but users are sensitive and want to know the applications they use to do their jobs are running on the best possible technology platforms and there's an obvious perception that things that are cheaper must not be as good as something more expensive.

Best possible SD-WAN experience

Building an SD-WAN is compelling because they offer the flexibility to leverage broadband, sometimes even consumer grade broadband to connect users to applications. Network engineers must ensure that neither network nor application performance is degraded in the process.

An SD-WAN can improve network and application performance by moving traffic to automatically route around points of network or link congestion. However, broadband is broadband and sometimes moving links doesn't address latency issues due to physical distance between sites.

WAN optimisation was designed to correct for errors but to also accelerate applications traffic so that users realise an improved experience and ultimately greater productivity.  From a workers' perspective, the evolution to an SD-WAN is something that will make staff to do their jobs with greater ease, efficiency and effectiveness.

So if you're a network professional considering evolving to an SD-WAN, do your career a favour and be sure to integrate WAN optimisation into the deployment. Users will have an experience that is noticeably better and will consider you to be the GOAT (greatest of all time) instead making you the goat.

By Zeus Kerravala, founder and principal analyst with ZK Research, for Silver Peak