Open Networking: The whale that swallowed SDN
- 18 June, 2014 06:49
The year is 2011, the Open Networking Foundation has just been formed by the largest Cloud and telecom service providers, software giants and academia – all of whom united at the inaugural Open Networking Summit under the common theme that the networking industry had fallen behind, its closed model stifled innovation and was not keeping up with the rest of the computer science discipline or the needs of emerging application and business requirements.
As you can imagine, this message was not warmly received by all of the networking incumbents who were squarely implicated. As a result, many responded by indicating that these SDN Cloud and software guys and academics didn't really understand networking, and that whatever technologies the incumbents already had were real SDN'. According to early incumbent rhetoric, OpenFlow was essentially a science project driven by idealists who didn't understand the real complexities of networking.
The SDN Laundromat
When OpenFlow first became popular, many framed the SDN debate as real SDN versus inferior and proprietary technologies. This seemed fair at the time as networking incumbents were positioning technologies that were clearly inferior and using aging development frameworks.
However, due to SDN's rapid rise in popularity, incumbents responded with significant R&D investments into alternative and complimentary SDN technologies, and by 2014 seem to have created a much better technical response to complaints they were simply re-framing existing technologies as SDN. For example, regardless of the positives or negatives of Cisco's SDN approach, there is no doubt that it has grown much more technologically sophisticated than their initial response to SDN in 2011.
Yet, at the 2014 Open Networking Summit, ONS Chair Guru Parulkar maintained his tone toward Cisco, including them in with "incumbents [that] have jumped in and are positioning other stuff they are doing as SDN," and further noting that "If you talk to people that really understand SDN, they will say that the Cisco approach doesn't provide the true value of SDN."
Re-framing the SDN debate
This apparent dichotomy has created a lot of confusion, with many observers feeling that if Cisco's 2011 SDN plan wasn't real SDN' because it was seen as a more antiquated technical approach, then why, as the solution has become more technologically sophisticated has it not become more accepted as a legitimate form of SDN by groups like the Open Networking Foundation.
To get to the center of this issue it is important to focus less on the what' of SDN the underlying technical components, and take a closer look at the why' of SDN.
Taking SDN back to its roots
During Martin Casado's speech at the inaugural Open Networking Summit, he noted that a primary value proposition of OpenFlow/SDN was simply the ability to innovate within the network.'
He recalled that the concept for OpenFlow began when working on systems that were critical to national security, with his frustration over the inability to freely code new solutions that would have solved critical challenges if networking products provided open development capabilities. Casado found that large numbers of organizations also shared in this desire to have the ability to innovate and differentiate in the network, forming the basis of the community that has helped OpenFlow grow to popularity.
Yet if we added a more open development environment to every switch in the market, it would be a big step forward, but would likely serve as little more than a temporary Band-Aid, never addressing the root cause of the problem.
To get to the deeper issues, we must ask just why it is that SDN has needed a huge counterculture movement just to get off the ground. Why did so many of the industries' largest consumers and most critical segments need to go to such great lengths to be heard, and what are the implications of this to the smaller and less well represented industry consumers? Why is it that networking's fork of the computer science discipline seems to have gotten so far from the best that computer science could offer?
These exact problems had already been faced during the transition from mainframe computers to mini's to today where computing hardware has become much more open, diverse and innovative.
The open computing model also caused the independent software market to explode by enabling anyone with an idea, ingenuity and drive to deliver new solutions that did not and could not exist under the closed and stifling conditions of the mainframe era. It is profoundly clear that closed mainframe-like conditions could not have resulted in today's open computer hardware and software markets which as a result have become the growth and innovation engine for the entire global economy.
An Industry with a Mainframe Mentality
Since well before the ONF was formed and even before OpenFlow was called OpenFlow, the core thrust of the entire SDN movement was never about promoting any specific SDN technology but has instead been about the larger goal of creating an open networking paradigm that would mirror the benefits of open computing.
To drive this point home, the director of Stanford's Clean Slate lab (where OpenFlow was born), Nick McKeown shared the following slides at the inaugural Open Networking Summit, highlighting the similarities between networking and legacy mainframes:
I am sure many of you who have followed SDN recognise this message as nothing new. Yet confusion around SDN remains rampant; industry conversations remain focused on competitive battles and lower-level technical details that don't have anything to do with creating a truly open ecosystem for networking and a core innovation engine for our industry that will keep it healthy, diverse and innovative going forward.
The New SD(lock)N
Today as many leading network technology providers are leveraging their immense power to drive support for their own closed SDN ecosystems, the industry and enterprises in particular stand at risk of being more locked-in to vendor technologies than ever. The burgeoning open and independent networking software ecosystem seems likely to be subverted into isolated pockets that will only work within vendor-specific ecosystems.
When server hardware is purchased, the owner is free to invest in software that is independent from hardware. I can purchase Microsoft exchange, and that does not create a business dependency on Dell Hardware or any specific hypervisor for example (Disclaimer I work for Dell, but the opinions expressed in this blog are my own and do not necessarily reflect the opinion of Dell). This paradigm is responsible for the emergence and massive growth of the independent software market and the vast sea of new business solutions that only became available as a direct result of the accessibility and economics created by open computing.
Yet if we look at emerging SDN and NFV ecosystems, it appears that application development will flow along vendor lines. Cisco and VMware for example are building comprehensive self-contained SDN/NFV ecosystems that are clearly separate and distinct. Unlike x86 where an investment in software does not create a dependency on an underlying hardware (or virtual hardware) provider, each ongoing investment into these ecosystems can create compounding dependencies on the platform provider, making it increasingly difficult for consumers to move between providers, diminishing their bargaining leverage and purchasing power.
Using OpenFlow or other open protocols does not inherently mean that these solutions will be any more open or interoperable than any legacy solutions. We can see emerging SDN ecosystems that leverage OpenFlow yet still require that applications be built specifically to work within a closed application ecosystem. Or in the case of companies like HP that have built customized OpenFlow-based ecosystems, extensive use of proprietary frameworks and custom extensions create the net effect of applications that will only work within vendor-specific ecosystems.
We Deserve Better
While many of us have come to believe that all this SDN stuff is just inherently more open, it seems clear that many leading enterprise SDN solutions will not deliver software that is free from dependence upon a specific physical/virtual infrastructure provider.
The deal that several leading SDN/NFV solutions offer is what looks like a step in the right direction. However this step would require consumers to invest heavily into closed vendor ecosystems and trust that there will hopefully be some answer to the lock-in and closed ecosystems eventually just trust us' echoes the common vendor refrain.
But the battle for the future of our industry has nothing to do with trust and everything to do with bargaining power. As we transition to the cloud era all of the rules of the networking industry are being rewritten. The new norms will be the result of vendor-centric Betamax wars along with battles in standards bodies and open source projects, and in all of these forums, vendors are well represented, huge technology firms are represented, but the average enterprise is not.
So how could enterprises be better represented? Groups like ONUG have emerged to give a voice to the finance vertical but for most industries there are no consortiums of this nature, leaving most enterprises isolated, fragmented and on the sidelines waiting as vendors chip away at the little bargaining leverage consumers are left with. The industry continues to be characterized by monopolistic conditions and it seems profoundly clear that hardware and software incumbents have stacked the deck heavily in their favor.
So what can we do?
When we look to see how open computing overcame vendor power to give greater leverage to the consumer, we see clearly that the clean abstraction between hardware and software was at the very core. This abstraction created powerful checks and balances that amplified the power of the consumer. Yet almost no networking vendors today are willing to offer this powerful choice.
An open networking model can empower businesses to invest into new technologies at a reduced risk a powerful strategic benefit which can accelerate business execution. This paradigm slices through the thick layers of vendor hype and confusing tactics, giving the ultimate bargaining chip to the consumer: If a vendor's software doesn't align with a business need, simply switch out the software leaving the considerable cost and complexity of the physical network, cable plant and topology in place, drastically reducing consumer exposure to the risks of new technologies.
Another key benefit of this model is that it allows consumers a way out of the Betamax wars happening today as vendors fight for control of future standards. These battles for vendor superiority offer little value to the consumer yet forces them to take on great risk and exposure.
Consider the recent battle between HD-DVD and Blu-ray; if we had a business need to invest in one technology before standardization, it would create significant exposure for our business. But, what if we could invest in hardware that could easily be converted between HD-DVD and Blue-ray it would significantly reduce the risk for our business to invest and drive business execution with new technologies. Open Networking allows you to do just this, providing standard hardware that can easily be converted between multiple different approaches to SDN and next-generation networking.
It's time for an Open Networking Revolution
While today it is clear that the SDN movement has caused significant inflection and change in the networking industry, confusion over the breadth, scope and complexity of the subject has created room for vendor FUD and confusing tactics that would leave consumers still beholden to vendors.
Today many leading SDN/NFV solutions promote closed application ecosystems that would move the industry from the legacy position where no real development could happen on a network device, to a position where each development made in networking software creates a compounding lock-in effect on the vendor platform. This sad attempt at compromise offers increased functionality in exchange for greater consumer lock-in and diminished bargaining leverage.
But it does not have to be this way.
The truth is, open networking is not the whale that swallowed SDN; SDN was never more than a piece of this much larger goal. But along the way we have allowed SDN to swallow open networking, allowing competitive battles over inconsequential technical components to distract us from the greater objective.
The internet and the future of networking are not just any old business deal, the internet has changed the world and will be the basis for an entire reinvention of our society. Watch a toddler's eyes up as they play with a tablet, watch how our children learn, consider how they will find work, how our society will find innovations to solve our great challenges and we see that the internet and networking technology is at the very cenre of our society's future. I for one do not want this future to be dictated and controlled by vendor influence and manipulated to support vendor goals. The future of our industry should not be beholden to vendor battles but should be guided by an open community where all of us work together to guide this critical foundation of society's future. It's time for us to raise our voices and demand better; it's time for an open networking revolution.