After years of design stability, how should businesses pivot to an IT infrastructure that will be subjected to almost continuous change?

Corporate wide area networks (WANs) used to be so predictable. Users sat at their desks. Information and software applications were stored in servers at company data centers.  And WAN design was a straightforward process of connecting offices to network hubs. This basic architecture served companies well for decades. 

Ho hum.   But growth in cloud and mobile usage are forcing information technology (IT) professionals to rethink network design.   Consider this: public cloud infrastructure will grow a whopping 36.8% this year, according to Gartner. Meanwhile, the enterprise mobility market is expected to double from 2016 to 2021.  This presents is a challenge for network architects.

Traditional WANs weren’t built for this.   Services like Multiprotocol Label Switch (MPLS) excel at providing fixed connections from edge sites to hubs.  But MPLS isn’t well-suited to changing traffic patterns.  Route adjustments are costly, and provisioning intervals can take months.

It’s clear that the move to cloud requires a fundamental shift in network design. Here are five recommendations for building an enterprise WAN that is flexible, easier to deploy and manage, and supports the high speed of digital change.

STEP 1: Build Regional Aggregation Nodes in Carrier-Neutral Data Centers.  The market is catching on that these sites serve as more than just interconnection hubs for networks and cloud providers.  In fact, colocation centers are ideal locations for companies to aggregate local traffic into regional hubs. The benefits are cost savings, performance and flexibility. With so many carriers to choose from, there’s more competition.  In one report, Forrester Research estimated 40% bandwidth savings when buying services at Equinix, one of the largest colocation companies. There’s also faster provisioning and greater flexibility to change networks if needed.  

STEP 2: Optimize the Core Network. Once aggregation sites are selected, they need to be connected. Many factors should weigh into this design, including estimated bandwidth requirements, traffic flows and growth. It’s particularly important to consider the performance demands of real-time applications.  For example, voice and video aren’t well-suited to packet-switched networks such as MPLS and Internet, where variable paths can inject jitter and impairments. Thus, networks carrying large volumes of VoIP and video conferencing may be better suited to private leased capacity or fixed-route low-latency networks such as Apcela. The beauty of the carrier-neutral model is that there will be a wide range of choices available to ensure the best solution.

STEP 3: Setup Direct Connections to Cloud Platforms.  As companies push more data to the cloud, the Internet’s “best-effort” service level become less suitable.  Direct connections to cloud providers offer greater speed, reliability and security. Many cloud platforms, including Amazon Web Services, Microsoft and Google offer direct access in the same carrier-neutral data centers described in Step 1. There is a caveat: It’s important to know where information is stored in the cloud.  If hundreds of miles separate the cloud provider’s servers from their direct connect location, it’s better to route traffic over the core network to an Internet gateway in closer proximity.

STEP 4: Implement SD-WAN to Improve Agility, Performance and Cost.  Software Defined WAN (SD-WAN) is a disruptive technology for telecom.  It is the glue that binds the architecture outlined above into a simple, more flexible network that evolves over time and is fully “cloud-ready”. With an intuitive graphical interface, SD-WAN administrators can adjust network parameters for individual applications with only a few clicks.  This means performance across a network can be fine-tuned in minutes, with no command-line interface entries required. Thanks to automated provisioning and a range of connection options that includes LTE and Internet, new sites can be added to the network in mere days.  Route optimization and application-level controls are especially useful as new cloud projects emerge and demands on the network change. 

STEP 5: Distribute Security and Internet Gateways.  The percentage of corporate traffic destined for the Internet is growing significantly, thanks in part to the adoption of cloud services.  Many corporate WANs manage Internet traffic today by funneling through a small number of secure firewalls located in company data centers.   This “hairpinning” often degrades internet performance for users who are located anywhere but the corporate data center. Some organizations instead choose to deploy firewalls at edge sites to improve Internet performance, but at considerable expense in hardware, software and security management.  The better solution is to deploy regional Internet security gateways inside aggregation nodes.   This places secure Internet connectivity at the core of the corporate WAN, and adjacent to the regional hubs of the Internet itself.  Costs are contained. Performance is improved.

The shortest path between two points is a straight line.  And the smaller we can make the line between users and information, the better their network performance will be.  By following these five steps, companies will be assured that their cloud-ready WAN will be an asset, not an obstacle.