Jim Frey is Vice President of Research, Network Management for Enterprise Management Associates (EMA) and is a featured guest blogger.
There were a lot of promises made when software-defined networking (SDN) first came onto the scene, and while some real benefits have been experienced, not all promises have been realized, or were even realistic. Such is commonly the case with new technologies: the initial hype exceeds the reality of the situation, but usually there are reasons to hold firm to the trend and take advantage of what the technology does deliver successfully.
While SDN concepts go back many years, the most recent formalized push began in earnest with the founding of the Open Networking Foundation in early 2011. With nearly four years of roadway behind us, now is a good time to check in on the hopes and promises of SDN to see which ones were real and which ones were not.
A good way to do that is to consider the status of the many use cases that have arisen around the technology by looking at some of the most popular cases.
Use Case #1: Save big money by commoditizing data center switching.
While some are enjoying real advantages, this is still a myth for others. The early hype was that OpenFlow-based “underlay” SDNs would turn the networking equipment world on its head by allowing all of the controls to be centralized into software controllers so that the actual switches themselves could be “dumb packet forwarders.” This has indeed been a huge boon to web-scale operations, such as Google’s poster-child implementation, where the traffic is fairly homogenous and there is an army of programmers available to build it and keep it running.
Service providers will enjoy similar success since they typically have the incentive and the resources to leverage such technology advances. But it has not inspired the majority of
mainstream enterprises, which continue to prefer stability and sophistication in their networking gear because they don’t have the resources or the risk tolerance to adopt a major architectural shift of this nature.
Use Case #2: Stretch virtual networks.
The “overlay” type of SDN appears to be the alternative reality of choice for mainstream enterprise. While tunneling and encapsulation predate the whole SDN movement by many, many years, a recent new crop of sophisticated choices for virtual network switching that can span Layer 2 domains and eliminate VLAN scale limits opens the door to the full promise of virtual server mobility. Further, these solutions can be implemented with little or no direct involvement by the networking team. The only myth remaining here is that they are right or ready for everyone. SDN overlays make the most sense for larger enterprises seeking to consolidate and leverage investments made in server virtualization and hybrid cloud infrastructures.
Use Case #3: Automate network functions and services.
This is the realm of network functions virtualization (NFV) and is made possible by the new levels of programmability and standardization that have come with SDN push. Service providers have latched onto this with a vengeance as they recognize an opportunity to expand service offerings while also stripping out both capital and operational costs. Enterprises understand the value too, but they look at NFV as a means to apply the same types of optimizations and controls with mixed hybrid infrastructures they have become reliant on in traditional, on-premise environments. While enterprises are lagging behind service providers in NFV adoption, this is clearly an eventual reality in both camps.
For the most part, these three core SDN use cases have ended up proving to be more reality than hype. EMA research published in mid-2014 found that while very few enterprises (less than 20%) had deployed SDN of any type, the overwhelming majority (over 80%!) was at some stage of research, evaluation, test, or deployment. Such focus and attention will certainly drive more use and more validation of SDN technologies in the months and years to come.