Fortune-500 companies frequently ask me about network transformation issues. They are concerned about Cloud applications having operational glitches and they want to hear how other enterprises are tackling Cloud-related networking issues. Current WAN architecture limitations create these problems, and most IT professionals use either backhauling or IP multicast techniques to keep their networks functioning. In order to understand why companies have this perception and why these problems persist, we must address the WAN legacy.
A Brief History of WAN
The precursor to the modern Internet, the ARPANET, was deployed in 1969, and with it the first iteration of WAN. From this, the Internet evolved, and from 1984 to 2004, four distinct generations of enterprise WAN technologies were deployed: time division multiplexing (TDM), frame relay, asynchronous transfer mode (ATM), and multi-protocol label switching (MPLS).
Unfortunately, we haven’t seen effective, new WAN technology since the mid-2000s. Modern enterprise networks still rely on two WAN services: MPLS and the Internet. Despite this lack of updated technology, enterprise applications shifted almost uniformly toward rapid Cloud adoption. How can the WAN keep up when application models are changing in revolutionary ways?
The Backhauling Challenge
Traditionally, companies design business WAN to include low-speed access to a service provider’s MPLS network at each branch office and to add one or more higher speed links at each data center. In this design, it’s common to backhaul a company’s Internet traffic to a data center before sending it through the Internet. This business WAN practice started when most data traffic was within one’s company data and flowed between branch office and data center.
In the 2015 WAN Architecture and Design Report, half of the IT professionals I interviewed indicated that they backhaul the majority of their Internet traffic. About 40% of them indicated that they backhaul more than 80% of their Internet traffic.
Backhauling enables a network organization to control traffic prior to handing it off to the Internet. However, it adds both cost and delay. The Internet traffic transits the MPLS link, which effectively increases overall public cloud services use.
Cloud providers try to improve application performance with global DNS, data center distribution, direct peering, and other options; however, the current WAN architecture is too rigid to adequately support any of these methods.
Booming public cloud services aren’t the only reason the WAN-design model is under attack. Cost reduction pressures, increased voice and video traffic requirements, and improved security and application performance are also contributing factors. Committed to solving these problems, vendors are launching a dynamic new architecture called software-defined WAN (SD-WAN).
We’ll continue to discuss where WAN is, where it’s headed, and what you should know about IP multicast in subsequent posts. Stay tuned!