Introducing the Software-Defined LAN: SDN at the Edge, and More
While SDN (Software-Defined Networking) is a justifiably hot topic in networking today, two obvious shortcomings are easily identifiable today. The first is that SDN is really all about protocols rather than operations-staff- and end-user-visible features, functions, and capabilities, and second is that SDN as presently defined and implemented has relatively little impact at the access layer – intermediary and edge switches and access points in particular.
Since these critical elements define not just the wireless LAN today, but the LAN overall, the concepts behind SDN - with appropriate implementations, of course - promise to have a big impact closer to the network edge. Welcome to the world of software-defined LANs (SD-LANs).
This series from Craig Mathias of Farpoint Group will examine the motivations, requirements, justifications, features, and benefits of the SD-LAN strategy, and why the SD-LAN will shortly be on the front burner at network shops everywhere.
Introducing the Software-Defined LAN: SDN at the Edge, and More
SDN has been a hot topic for several years now, and justifiably so. The whole idea is to make networks “softer” - more flexible, adaptable, and efficient, as well as to take advantage of advances in off-the-shelf hardware and enable customization to meet whatever needs might materialize in the field now and in the future.
SDN, then, is more than a good idea – it points the way to the future of networking itself, shifting in a very productive manner how we think about, plan, implement, manage, and use networks. If you’d like to dig in deeper, you can find a significant amount of background material on SDN here.
But while this flexibility enables policy-driven adaptability, a faster response to changing conditions, enhanced security, simplified fixes, and other improvements as may be required and desired, and many additional benefits, it’s quite fair to look at SDN as really more of a data-center solution than a network-wide technology. After all, what is SDN?
While even (or perhaps, especially!) the analyst community argues over the specifics, working definitions almost always come down to a discussion of protocols (most notably OpenFlow), which is, after all, to be expected in a field that defines itself largely based on protocols.
Don’t get me wrong, protocols are terrific and essential, but they are clearly only one piece of the puzzle that is the future of networking itself. SDN is thus a good start, but we really don’t have complete picture of where networking is going from a consideration of SDN alone.
Expanding on this theme, since SDN is mostly about the data-center core of the network, how are the benefits of software definition to be extended to the edge of the network, particularly to the intermediate and edges switch and access points (APs) where end-user demand hits the network and all it connects to?
After all, if SDN is such a good idea (and it is), it makes sense that the benefits of SDN should, and indeed must, be realizable out where demand is generated, and end-user requirements are satisfied.
With this as a point of departure, it very quickly becomes clear that a discussion of protocols and functional elements alone isn’t going to provide much motivation to IT managers struggling with the best way to get an optimal return on scarce resources, building sustainable solutions that reduce not just capital expense, but especially labor-intensive operating expense – the cost of the people who design, operate, manage, and fix networks in production environments.
As I’ve said many times, advances in networking are best evaluated in terms of improvements not just in the traditional metrics of throughput and capacity, but equally-importantly in terms of operations-staff and end-user productivity. And here, I’m afraid, SDN is not yet ready for prime time.
SD-WANs and SD-LANs – picking up where SDN leaves off
But all is not lost – in fact we’re just getting started here. Consider, for example, the software-defined WAN (SD-WAN), now gaining popularity with the carrier community. Like SDN, SD-WAN separates the control and data planes of the WAN, and enables a degree of control across multiple WAN elements (both physical and virtual) not otherwise possible.
The key point here is that a portion of the WAN community has recognized the need to software-define the WAN in a manner not necessarily identical to the protocols that today define SDN – so they’re related, but different due to essentially different requirements.
So, then – can we apply this same thinking to the LAN? Does SD-LAN – extending software-defined techniques into the access layer of switches, APs, and perhaps even, via drivers, eventually even into client devices, make sense?
I believe it does, but to so conclude we must understand two key elements: What requirements must be placed on an SD-LAN implementation, and how does softening closer to the edge benefit network managers and end-users? In other words, features, benefits, and advantages – same as it’s ever been.
Our next two articles in this series will examine each of these in turn.
One other very important point here: Returning to a theme I’ve examined in detail over the years, success in SD-LAN ultimately depends to a very large degree on the architecture of a given WLAN system solution. Note: This is independent of additional innovation on the part of the IEEE 802.11 folks and the efforts of the Wi-Fi Alliance – these have become the jacks-or-better baseline for any successful player in wireless LANs going forward.
We’ve traditionally thought of architecture as benefiting the overall performance characteristics of a given installation (much more on this next time), but now we’re asking it to contribute to advances in capacity, reliability, operations productivity, security, scalability, total cost of ownership, and much more. And with wireless-first a growing reality and requirement, the WLAN industry is thus continually challenged to provide more-than-meaningful advances in much more than throughput alone.
The expectations surrounding SD-LAN, then, are quite serious.
SDN, no argument, is pretty exciting. SD-LAN, I think you will shortly agree, is pretty exciting, too.
All posts in this series
- Introducing the Software-Defined LAN: SDN at the Edge, and More (1 of 3 in a series by Craig Mathias)
- What Are The Requirements For SD-LAN? (2 of 3 in a series by Craig Mathias)
- What Are The Benefits of SD-LAN? (3 of 3 in a series by Craig Mathias)
For More Info on SDN, Read This Series of Articles By Marko Tisler
For More Info on SDN, Read This Series of Articles By Craig Mathais