Five Ways the Enterprise Network Will Change in 2014 and Beyond

Data is the lifeblood of the enterprise, networking its circulatory system. And the changes set to take place in enterprise networking in 2014 mark nothing less than the evolution of an entirely new species of infrastructure.

This creature will be more efficient and more effective at linking users, applications and data, encompassing the abilities not only to connect resources across increasingly distributed architectures, but to self-manage and self-govern itself to such a high degree that human oversight will likely be restricted to only the highest-order policy management.

Many of the tools leading to this new environment were introduced in the past year or so, and the next 12 months will see their implementation in real production environments, ushering in the beginning stages of a near-total transformation of the networking stack over the next decade.

Here then, are the key components of this new creature, all of which either directly support or are a subset of the first entry:

Software Defined Networking (SDN)

The separation of the control and data plane is one of those things that is both revolutionary and yet so obvious that it’s amazing it didn’t happen years ago. The ability to define and manipulate network architectures completely in software will give the enterprise the unprecedented capacity to tailor complete network environments, down to the application and even the user level. Combined with virtual server and storage architectures, knowledge workers will no longer be restricted by static enterprise infrastructure. Instead, they’ll be able to self-provision their own end-to-end data environments using the exact mix of resources that suits their needs. At the same time, the enterprise should see its network operating costs drop significantly as a result of this self-defining, self-correcting network.

Open Vs. Proprietary

About the only major question surrounding SDN deployment now is whether to go with an open platform like VMware’s NSX or a more hardware-integrated approach, such as those espoused by Cisco and Juniper. Theoretically, an open platform provides for greater hardware flexibility, although some would argue that with VMware calling the shots, it merely results in single-vendor dependency on the virtual layer. The integrated approach, however, would embed key SDN functionality on the network processor, which requires a hardware upgrade to get started but in the end promises a more streamlined, cost-effective architecture. Expect this debate to be one of the key decision points in the SDN deployment phase.

Goodbye Spanning Tree, Hello Fabrics

No matter how this plays out, it is clear that the traditional Spanning Tree architecture will fall to more flexible fabric topologies over the coming year. When the time comes for live virtual machine migration and linked Ethernet connections, the converged network fabric is the only way to go. The key component here is the data center bridge, which at the moment seems likely to incorporate the Transparent Interconnection of Lots of Links (TRILL) protocol. The top challenge will be to build the fabric with enough resilience and flexibility to meet the needs of software defined architectures while maintaining both multipath connectivity and stability within individual links.

Network Intelligence

This, of course, leads us to the rise of network intelligence. The human brain is simply not capable of carrying out the hands-on management of the emerging dynamic network. As advanced automation and orchestration become the order of the day, network elements will need to adjust to the shifting data loads and growing diversity of this new data environment. Intelligent systems are already making their way to the edge in order to parse incoming traffic from mobile users, machine-to-machine communications and the Internet of Things. With the ability to more accurately gauge the types of data traversing the network, intelligent systems will be better able to avoid bottlenecks and reduce both the cost and complexity of the data environment.

Optical Networking

Optical links are usually associated with long-haul carrier networks, but as the data center becomes larger and more complex, glass starts to make more sense, particularly on the LAN. With optical links, you not only get high speeds and more bandwidth, but you can also push switches and other devices to the edge, providing a flatter network topology more amenable to fabric architectures. Additionally, single-mode fiber is becoming increasingly cost-competitive with copper, making it difficult to justify electronic communications on budgetary terms.

To be sure, a host of innovative new solutions will emerge up and down the stack in the coming year, but this list represents the major trend lines leading to a new enterprise network. Some may argue that I left out infrastructure consolidation and modularity, which have the potential of pushing the network onto the PCIe bus or even the memory interface. But while that may prove to be the solution of choice for cloud-facing, hyperscale facilities, it seems the typical enterprise is likely to leverage its discrete network infrastructure, at least for now.

But make no mistake, that network is changing in a fundamental way. And in a few short years, it will be all but unrecognizable to today’s network architect.

Photo courtesy of Shutterstock.

Latest Articles

Follow Us On Social Media

Explore More