InterOp has always served as a bellwether for the industry. Year in and year out, the InterOp conference agenda and exhibit floor highlight what is hot. And, by omission, what is not.
These days, SDN is hot – and getting hotter. This means InterOp attendees must be ready for an SDN onslaught in Las Vegas this year. While it will be encouraging to see so much engineering and marketing effort focused on SDN advancements, separating the wheat from the chaff will be a significant challenge for even the most astute observer.
As I head to InterOp next month, I’ll be on the lookout for certain things that I believe serve to move SDN forward… faster. Here’s my top 10 list:
10. More Weight Behind Fewer APIs
Network programmability is core and critical to SDN success. APIs serve as the primary programming mechanism among SDN controllers, devices, services, applications, and management systems. The problem is, right now, we have too many API options operating at all levels of the SDN structure. What we need is a smaller and standardized set of APIs that narrows the focus for developers. This focus will accelerate SDN service developments, solution availability, and, most importantly, operator deployments. At InterOp, watch for momentum around some core set of SDN APIs.
9. More Usability Improvements
It could be argued that usability (or lack thereof) is the primary driving force behind SDN. Core SDN advancements focused on network consolidation, automation, and virtualization drive big gains in deploying, operating, and enhancing not only the network, but also the computing infrastructure (e.g., VM mobility) and cloud services (e.g., self-service). At InterOp, watch for SDN management systems that take more action, while requiring less operator interaction.
8. More Analytics Included in SDN Solutions
Given the dynamic nature of the SDN environment, it is imperative that network state serves as a primary input to not only precise network forwarding, but also optimal network programming. Some early SDN solutions (e.g., Open vSwitch, Lyatiss, Plexxi) point to increasing support for integrated analytics. At InterOp, watch for real-time analytics moving from reactive discrete management system to proactive integral feedback mechanism.
7. More Step-by-Step Guidance for Operators
SDN is often presented as the ideal destination for networks – and network operators. Unfortunately, little to no specific direction is provided to operators mapping out their SDN futures. For SDN to be more evolution than revolution, deployments must be incremental. Defining these increments is the responsibility of SDN suppliers. At InterOp, watch for suppliers specifying small sure steps that lead to eventual and, hopefully, inevitable SDN success.
6. More Mainstream Operator Deployments
While I applaud Google, Microsoft, Stanford, and all the other leading edge operators for providing early views into SDN capabilities and benefits, mainstream enterprise deployments indicate that SDN can serve all networks – not just those with extraordinary requirements and resources. At InterOp, watch for SDN solutions deployed outside of high-tech and education – e.g., financial, retail, government, and manufacturing.
5. More Solutions Aimed Outside the Data Center
The data center has been the early focus of most early SDN discussions and solutions. Given the many pressures put on the data center, this is understandable. However, considering the cost and complexity of operating the broader enterprise or service provider network, SDN offers many advantages to such segments as LANs, WANs, branch networks, and cloud services. At InterOp, watch for SDN solutions aimed at the enterprise and service provider WAN (e.g., optical networks, data/content delivery, rapid/remote provisioning, NaaS).
4. More Momentum Behind Fewer SDN Controllers
As with APIs, too many SDN controllers slow SDN developments and deployments. While freedom of choice is the core of the open networking movement – a movement closely tied to SDN – too many controller choices serve to confuse operators and dilute developer efforts. Natural industry selection of two or three SDN controllers drives more focused development above (e.g., network virtualization, data acceleration) and aside (e.g., analytics, automation) the network core. At InterOp, watch for building momentum around select SDN controllers.
3. More SDN-Focused Support Services
With SDN, it is not to be business as usual for network support services. Increased emphasis must be placed on systems integration, software management, computing extensions, application development, and network optimization. SDN is not your father’s network. No longer will smooth installation and rapid replacement serve as the primary indicators of support service excellence. At InterOp, watch for “first-step” support services focused on legacy migration and application validation.
2. More Platform Optimization for SDN Software
SDN drives software functionality out from behind the curtain (i.e., device operating system) to center stage (e.g., controller, SDN services and applications, automation and analytic systems). While the flexibility and functionality of such software-based systems will be most welcome in networking, old perceptions of slowed performance and suspect reliability must be proven incorrect in this brave new SDN world. At InterOp, watch for a tighter coupling of SDN software systems to developments in processors, servers, networking devices, service appliance/adapters, and even “host” software systems (e.g., orchestrators and hypervisors).
1. More Timelines, Fewer Teasers
Grand visions and glorious architectures have dominated SDN supplier announcements for two years now. It’s time to turn the vision into reality. From now on, the expectation is that SDN announcements should include proven products and firm delivery dates. With that said, operators shouldn’t expect too much in every announcement. Remember, SDN is more evolution than revolution. As such, new solutions should provide incremental improvements to the status quo. At InterOp, watch for working available solutions focused on certain network problems (VM mobility, data delivery, and traffic control) and segments (e.g., data center, WAN, and LAN).
Am I expecting too much from my time at InterOp? Will I leave Las Vegas disappointed? Or will I be pleasantly surprised? What are your expectations for SDN at InterOp? I’m sure I’m not the only one interested. Enterprise operators, networking vendors, service providers, software developers, systems integrators…all are trying to figure out where and when SDN hits the world’s networks – and all will be at InterOp listening and learning.