The network of the future

Software-defined WANs will change the status quo of networking.

Stuart Hardy
By Stuart Hardy, business development director of EOH Global Networks Division UK.
Johannesburg, 19 Jul 2016

In my last Industry Insight, I looked at how MPLS is dead as a result of users consuming all business applications from a cloud infrastructure or software as a service platform. The technology that will change the status quo will be software-defined WANs (SD WANs).

It makes a lot of sense to want to use the Internet as an access medium to a new future network. In domestic networks, the quality is good, cost is low, and it offers considerably more bandwidth.

But, when companies try to use the Internet in global applications or WAN delivery, things slow down and performance drops by up to 99%.

Aside from the Internet, the other starting point for this network would be the cloud. After all, all applications will one day be delivered from the cloud. So integrating into every major provider would be essential, as the centre point of the network needs to be the application.

Users know that MPLS networks are the best to deliver applications as they have lower latency and zero packet loss. These elements are essential to good and consistent application delivery. So, a network is needed that allows users to connect to it using a domestic Internet connection, but which is intelligent enough to route business-critical and sensitive applications locally through a low latency global L2 or MPLS core.

Eliminating complications

To resolve the issues being caused by latency in the user experience, application optimisation is also necessary. More specifically, TCP IP acceleration and the rest of the optimisation stack, such as caching, compression and deduplication. But, instead of having to overlay this on top of an MPLS and Internet network, it would make more sense to embed this capability into the POP network at a carrier level. That way, customers can get optimised application access automatically and remove the growing complexity of building a layered network.

The benefits of this SDN would be compelling:
* Because the network is software-based, users can connect to it in hours or days as opposed to months.
* Because it leverages the Internet, network architectures will be simplified, as there will only be a connection to the Internet.
* The intelligence in the network would identify business-critical applications coming from the connection, and divert them locally over a global dedicated L2 or MPLS core network to ensure the best application and user performance.
* Because the network already integrates with every IaaS, SaaS and PaaS on every continent, users can get immediate access to their applications, fully optimised end-to-end, just by adding the application IP address and URL.
* And, if all of that is not exciting enough, the TCO of this network in comparison to layered networks would be 50% less in most cases.

The SD WAN is an exciting space and offers serious benefits.

Managing a WAN through software provides many advantages, including the ability to be notified of any issues and being able to manage the entire WAN through a single interface. In the past, making changes to network configurations would have required manual configurations being created and installed, as well as an on-site technician to do this. With SD-WAN, the entire WAN can be controlled, managed and changed easily.

While this may seem like an impossible wish list, there are two companies in the market today that are doing pretty much this: Aryaka and Virtela. The major difference between these SD WAN companies and the rest is that they own the underlying network that connects their global POPs, and they don't rely on the Internet to deliver global applications. The result is considerably faster application and WAN performance.

The SD WAN is an exciting space and offers serious benefits. It's also a big change for a lot of companies, which makes IT managers nervous. So test them. Unlike MPLS, companies don't have to sign a three-year agreement and deal with a four-month roll-out to then see if they achieve the desired results. These SD WAN networks can be deployed where a company's biggest issues are, and results can be measured in days.

Either way, I believe SD WAN will replace traditional networking, and one day businesses will most likely buy their WAN from a company they have never even heard of, one that does not own any fibre in the ground.