Virtualization, the cloud, software-defined networking (SDN), and myriad other forces are driving tremendous changes in enterprise networking. These changes are usually measured by their effects on the delivery of applications and services, but the impact on physical infrastructure deserves a closer look, too.
In the router market, for example, the cloud is shifting the demand in branch offices from low-end solutions to mid-range, according to research firm Infonetics, which reported an eight percent jump in mid-range router sales in the fourth quarter of 2012. As service-based environments start to take hold, workers across highly distributed architecture will require the same connectivity that those at headquarters enjoy. In fact, this was one of the few bright spots for routers in 2012, which saw the market fall three percent from 2011.
Meanwhile, leading vendors are hoping to keep their hardware platforms relevant now that advanced SDN architectures threaten to push everything onto commodity devices. Cisco has a new Integrated Services Router (ISR), the ISR-AX, which packs more intelligence under the hood to provide a high degree of “application awareness” and advanced delivery capabilities in highly dynamic environments. The goal is nothing less than to turn the lowly router from a simple packet forwarding device to a full application enhancement platform, as Inbar Lasser-Raab, Cisco's senior director for mobility, routing and WAAS, told Enterprise Networking Planet's Sean Michael Kerner.
But even while advanced configuration and delivery capabilities increase, so too does the need to lower capital and operating costs. Serving these twin masters is the primary goal of Juniper’s new PTX3000 core router, which packs 3.80 Tbps capability into a 10-inch box. The device has an 8-slot chassis capable of forwarding 240G per slot, and with a new set of line cards forthcoming, Juniper is promising upwards of 24 Tbps per box before long. And with either 24 10 GbE or two 100 GbE interfaces, enterprises can establish core capabilities at a fraction of the cost and space of current designs.
And since full functionality in the cloud depends very much on open network environments, management stacks are starting to embrace the new reality. Mellanox, for example, is leveraging its experience with InfiniBand to bring greater openness to Ethernet in its MLNX-OS platform. At the moment, the company is looking to open up Layer 2 and 3 functions through GPL and BSD licenses, which would give enterprises the ability to customize their network functionality while still retaining compatibility with the wider data universe. In the old days, this wasn’t much of a concern, because enterprise networks were essentially closed systems to begin with. But with hyperscale environments becoming the norm, connectivity must be as broad as possible to ensure the highest degree of flexibility.
All of this activity brings an ironic twist to modern network architectures in that control of the physical layer, currently one of the biggest burdens facing the enterprise, will become easier even as new logical architectures become more complex. With new platforms providing tools like dynamic load balancing and deep-dive monitoring and analytics, circumventing bottlenecks and other performance-diminishing hazards should be a snap, and in fact will likely become fully automated over time. Defective hardware, meanwhile, can be decommissioned at a moment’s notice and then swapped out at a leisurely pace while traffic is rerouted to functioning devices. No more frantic calls in the middle of the night because key network pathways are blocked.
It almost sounds too easy, but that seems to be the world we’re heading into – a new network for a new age of digital communication.