Monday, April 9, 2012

Why NEW architecture will happen


When economics combines with cloud performance on the WAN, the NEW architecture will gain popularity.

In my first column, I described a Next-generation Enterprise WAN (“NEW” for short) architecture. Here I’d like to cover why this NEW architecture will be widely deployed over the next few years.
Part of the reason – and some would argue, the primary reason – is, to paraphrase Bill Clinton and James Carville, "It's the economics, stupid!"
MPLS, from telcos like AT&T, Verizon and BT, is the dominant enterprise private WAN architecture in the U.S. and worldwide, and put simply, it’s very expensive. Branch and mid-sized office copper MPLS connectivity for U.S. locations typically costs $300 to $600 per Mbps per month, versus broadband Internet connections, which run $1.50 to $15 per Mbps per month. Fiber-based MPLS data center connectivity is typically in the $60 to $200 per Mbps per month range, versus Internet bandwidth costs at a carrier-neutral colocation facility of $10 to $20 per month.

That’s an order-of-magnitude price difference between MPLS and Internet connectivity at data centers and an almost two orders-of-magnitude price difference for branch connectivity. In future columns, we’ll cover why this price gap is so enormous – and likely to remain so “forever” – but for now it’s simply enough to note the reality. Since no organization’s WAN budget is growing as fast as the data is on their networks, pretty much everyone will need to leverage public Internet connections to augment or even replace their expensive MPLS circuits.
But if your analogy preferences run closer to light beer commercials than Democratic presidential campaign slogans, then maybe a better description would be: “Tastes Great” and “Less Filling.” Here, “Less Filling” is the aforementioned price/performance economics, while “Tastes Great” is the added reliability and application performance predictability the NEW architecture delivers.
WAN Virtualization technology delivers this network reliability and performance predictability, wrapping a layer of software intelligence around multiple public or private WAN connections in much the same way RAID wrapped intelligence around cheap PC hard disk technology and revolutionized storage 20 years ago. And, in fact, it is this lack of performance predictability, which is why, even as everything else in IT has moved to leverage the public Internet, the enterprise WAN itself has remained private rather than migrating to public-Internet IPSec VPNs.
But in addition to enabling Internet economics, reducing monthly dollar OpEx, the performance predictability delivered by the WAN Virtualization (plus typically WAN Optimization and distributed replicated file services) technology of the NEW architecture has additional benefits as well.
The four technologies mentioned, combined with the small footprint enabled by server virtualization, will enable a smooth migration to cloud services. Without reliability and performance predictability, most IT folks simply will not be willing to move applications to a private cloud, to say nothing of hybrid or public clouds. Having a private cloud at a colo facility connected by a predictable reliable WAN under enterprise control in particular will be the way that the vast bulk of enterprise applications can safely and securely be migrated to take advantage of public cloud services, at a pace that makes sense for each enterprise’s computing department, without blowing the budget nor causing users to complain about unusable applications.
This NEW architecture reduces network management costs – people OpEx – by reducing troubleshooting costs and virtually eliminating MTTR issues. Centralizing network and IT complexity for things like Internet access/intrusion prevention and remote site backup saves time and money while also improving user experience.
The NEW architecture also enables the necessary additional bandwidth and performance predictability for newer, network-stressing applications like HD videoconferencing and VDI/DaaS, which MPLS cannot reasonably deliver in a cost-effective manner for most organizations.
A final reason this NEW architecture will become common is that it does not entail a forklift upgrade, but rather can be added incrementally to an existing enterprise WAN. The deployment of WAN Optimization technology has shown the deployment benefits of such an incrementalist approach. And by safely and securely enabling IT migration to public and hybrid cloud computing, it will be a key accelerating mechanism for that migration as well.
All of the above reasons explain why a NEW architecture is needed; which one or which combination prove to be the most important, as well as why the NEW architecture ends up being adopted, are factors we’ll only know for sure in a few years.
What we do know is that betting against order-of-magnitude economic leaps is rarely a good idea, and in tech infrastructure, the combination of superior economics and superior reliability pretty much always wins.
And when “It’s the economics, stupid” meets up with the virtuous cycle interrelationship of “It’s the way to make cloud computing work on my WAN when and how we want to do it,” you’ve got a once in a decade NEW architecture that will give early adopters significant competitive advantages, and likely make a few careers as well.
A leading expert in WAN/LAN switching and routing, Andy founded Talari Networks, a pioneer in WAN Virtualization technology, and served as its first CEO. Andy is the author of an upcoming book on Next-generation Enterprise WANs.
Post a Comment