The year is coming to a close. While the notable innovations and riveting conversations that helped move the software-defined networking (SDN) industry forward in 2014 are worthy of reflection, I can't help but think of all the exciting changes that lie ahead in 2015.
Prediction No. 1: Open-source software will become the new norm for network standards
People in networking have decades of experience with network standards -- in committees. Whenever a new feature or capability has been required, a committee has been formed (or repurposed) and a standard has been hammered out in the form of a specification written on paper. And at some point during the process, companies begin to implement the standard. In the past, implementation has required a serious commitment to hardware, with its typical attendant cycle times of one to three years. So it's always been dangerous to get too far ahead of the standard, lest a change is instituted before it's finalized.
With SDN, the development of standards by committees (called de jure standards) applies mainly to the hardware, but not the software. Hardware becomes more general purpose and still relies on transmission and switching standards, like WDM and Ethernet framing. But the software that controls it no longer resides in the hardware. Rather, it resides in general-purpose servers in a data center or cloud facility. Much of what the software performs does not need to be standardized; usually only some interfaces do. Software interfaces in servers are almost always better standardized in a de facto manner by open-source communities.
The Open Networking Foundation (ONF) strongly supports the open-source approach. We have expended considerable efforts on traditional standardization -- for the OpenFlow substrate -- because it is a protocol on a wire between products made by different manufacturers and is often implemented in hardware. This we will continue. But when our attention moves to the software layers in the server farm, we have supported and contributed open-source software.
In 2015, I predict that open-source software will be recognized not only as a legitimate route, but the desirable route to network standards. Vendors will look to open-source software as a way to reduce development expenses on things that don't meaningfully differentiate products. Network operators will begin adopting open-source software directly or indirectly, or by starting a project themselves and sharing it with the community for further development. Old networking standards folks might find this unnatural, but by now we should all know that technology moves quickly, and we need to evolve or become extinct.
Prediction No. 2: Network operators will demand open SDN, not vendor SDN
This prediction has two important aspects. First, network operators will recognize the value of SDN, not just network virtualization or Network Functions Virtualization (NFV), neither of which can reach its potential value without SDN underneath. Secondly, network operators, including some enterprises, will start to see through vendor products touted as SDN that in reality retain vendor lock-in and proprietary protocols and interfaces, and in many cases do not even include the physical separation of forwarding and control. When as many operators as today are deploying new, green-field networks using SDN, they have the perfect opportunity to demand truly open SDN, and demand they will.
The open and interoperable solutions and products that ONF encourages protect users from vendor lock-in, offering business benefits including capex savings and the ability for companies to select products based on their needs, rather than being tied to a particular company's offerings. Through the establishment of our OpenFlow Conformance Testing Program and the experiences of companies that have completed testing, I've seen conformant products shift from a bonus to a necessity. Openness is becoming a deciding factor for end users as they migrate to SDN.
Prediction No. 3: New OpenFlow products will make it the default southbound choice
I've been hearing some people say that OpenFlow is past its prime, that "it's not available, and we don't need it anyway." This is only wishful thinking by vendors that want to deprive network operators of an open ecosystem of commodity packet-processing hardware and software. Great hardware support of OpenFlow beyond v1.0 has been rather long in coming, but it is coming. Between the notable support of OpenFlow v1.3 and the fierce race among chip industry leaders and challengers to prove who has the best support on fixed ASIC, flexible match-action ASIC, NPU, FPGA, or even pure CPU platforms, soon we can stop jumping through hoops to avoid using OpenFlow.
The most interesting part of SDN for network operators is the application, orchestration, NFV, and policy software that brings direct business value from the network -- but OpenFlow is the enabler of all that value. It's the transport that connects the thinking (the apps) to the doing (the packet processing).
Prediction No. 4: Skills training will emerge as the biggest SDN growth area
Sure, IDC says that the compound annual growth rate for 2014-2018 for SDN in the enterprise and cloud data center will be 89.4%. But organizational process changes to adapt to SDN will take time to implement. And the two main concerns about SDN focus on security: network security and job security. There have been concerns about the changing role of the IT professional in the automated network since SDN's conceptual introduction. Things are changing. The knowledge, skills, and abilities required throughout and following this transition will skew toward software rather than hardware. So in 2015, skills training will kick into gear.
What are your predictions for the SDN industry in 2015? Share your thoughts in the comments below!