We believe that working together in the open is the best way to solve a complex set of problems. We do not want to solve things repeatedly on multiple overlapping fronts. We favor working collaboratively to solve a challenge one time, in the most effective manner possible. New pursuits should deliver value in new areas or should distinctively improve existing solutions. Those of us working on Open Compute would assert that this is the best way to innovate on these hardware platforms.

An example of this is a partner of ours who has a very effective fabric interconnect solution, but does not engineer a power supply; they simply consume an off-the-shelf power supply and have integrated it into their product. They have some concerns about how to optimize or tune that power supply subsystem, but their focus on the fabric is what really distinguishes them.

Our belief is that Open Compute can fill the gap where such design problems exist. In the example above, the project encourages work on the power supply problem to proceed through the community with specialists who really distinguish themselves in that arena, while the partner with interconnectivity fabric expertise can acutely focus on its own area of prowess. Open Compute keeps one from working in a single silo across the entire stack, and makes development more like a pyramid where work builds upon on a strong and open base and provides opportunities for more focused and distinctive innovation at the pinnacle.

The Open Compute Project needs requirements, perspectives and inputs to help us ensure that we are solving the right problems. We’ve seen providers willing to open source attributes of their computational fluid dynamics software, and other providers open sourcing elements of their CAD or eCAD software to enable development of electrical or mechanical designs by the community. As activity and involvement escalate within the project, the community hopes to see more of these things become available to open up new opportunities for involvement.

The Open Compute Project has a high degree of synergy with OpenStack because it ultimately becomes the foundation, or the platform, on which the operating system of the cloud can run. In very similar fashion, Open Compute has a large community of people working together to optimize and define what is needed in data centers and hardware platforms. When you have multiple people working together in the open, you can accomplish great things.

Together, we think that OpenStack and Open Compute represent a complementary utopia for the cloud computing industry. In this utopia, even competitive parties can collaborate and find opportunities to deliver value across the entire solution stack. I often say this means we are driving to be truly open “from the gutter to ground.” This spirit of openness and collaboration can take us to exciting new places that we have not yet be able to go before.

A perfect 1.0 PUE would mean that all the power delivered to the facility is available for servers to consume. Typically, power is lost in conversion, transport and delivery from a power utility to a server system. A common PUE you will see in the industry ranges between 1.2 and 1.5. Facebook has been able to achieve a PUE that is sub 1.1. As noted this is an impressive accomplishment.

Another area to which we look for advancements in sustainability is simplification of the servers themselves. Focusing on things like material selection and sheet yield, reduction in component count and optimization of the overall assembly process can yield real environmental benefits. All of these things play into a very effective environmental envelope meant to deliver solutions optimized for environmental stewardship from the inside out.

We also have the distinction here at Rackspace of leading the Virtual I/O community within Open Compute. This is geared around defining a framework that provides opportunities for cost reduction and flexibility not traditionally available in server platforms.

Our initial goal in the Virtual I/O project is to develop solutions that provide modularity and flexibility in assignment of expansion devices such as storage and network adapters to platforms. We often describe such elements as key attributes of the “personality” of a system. This “personality” includes devices such as network interface cards, storage resources and the like. With Virtual I/O, we want to make these things programmatically assignable so that we can define the capabilities of customer systems “just-in-time” rather than rush about the data center plugging in and managing individual components.

I have to drop back a few weeks to the OCP Summit to best set the scene for these reflections. In this pursuit, I’ll give a quick run-down of some of the memorable and foundational elements shaping my present perspective. Here, I should also say that while I am not hoping to provide a comprehensive or journalistically sound portrayal, I can catalog a few first hand experiences that distinguished the event and many subsequent engagements for me.

During summit registration on May 2nd I was able to greet many colleagues and counterparts from throughout the world in the newly constructed lobby and museum area of our San Antonio home, dubbed The Castle. There was plenty of very positive feedback on our facility and visitors had the chance to absorb some of the cultural accouterments that exemplify the character of Rackers (Rackspace employees). Even now, these thoughts strike me with feelings of pride. As a Racker, such feelings are generally not anomalous, but the opportunity to receive the compliments and positive commentary of so many others resonates still with a memorable frequency for me.

During the keynote session that followed registration, Frank Frankovsky introduced the why, what and how of Open Compute. His presentation was excellent. In reference to his portrayal of the functional and philosophical tenets of OCP I am in fact hesitant to say “that was exactly what I’ve been saying,” but I was struck repeatedly by this point. He emphasized and well expressed our commitment to collaboration, openness, innovation and efficiency and set the scene for the introduction of new partnerships and projects.

Others, including Rackspace President Lew Moorman, had the chance to express their perspectives and commitment to the project, and new partnerships and contributions were announced. Here I was struck with an odd intermingling of pride and humility. We are making real progress on these projects. Whether we look to Open Rack, Open Vault, Hardware Management, Motherboard design or Virtual IO, there are plenty of people thankfully more talented than me doing really amazing work. We got to show the world the Open Rack R2D2-sized prototype, and here I can say that the efforts of Facebook mechanical engineers like Jon Ehlen clearly illustrate the statements Frank asserted concerning the complexity of solving problems in an elegant and vanity free fashion. I’ll do later posts with more detail on other projects including Open Rack, but let it suffice to say that production solutions of broad and impactful applicability are near.

We rounded out the remainder of the first and the majority of the second and final days with collaborative design sessions. I’ll talk more about the dynamics and mechanics of these as I survey the individual projects in detail, but the Virtual IO sessions led by my friend and long time co-Racker James Hesketh precipitated plenty of debate and positive interaction. (I joined these sessions as stenographer, taking embarrassing notes with comments including “guy in grey shirt says.” I’ll apologize to those of you I didn’t recognize. I’m looking forward to  grabbing a beer with each of you as we get you in, introduced and involved – “guy in grey shirt” definitely included).

Now I realize that I’ve gotten long winded and avoided expressing much about my new friends and experiences here in Taiwan. Again, more to come on this topic, but I learned a lot. People are changing the way they do business and while OCP is clearly not solely responsible, it plugs in to this reformulating landscape with an opportunistic and aligned synergy. To me, this is a synergy of the moment and is as much about OCP being fortuitously positioned in time among other developments as it is about OCP reshaping or defining anything. Maybe that really is the why and how of what will be our success. No one thing will change the world, the world simply changes. What we need are catalysts to shape and guide the change in a positive fashion.  My realization is that catalytic strategies and individuals are far ranging, present and accounted for. I now have a wholly new aspirational perspective on the pace, scope and opportunity that is within reach.

The last thing I have time to say is this: You can say we here too. Why would you not want to? All you have to do is hold up your hand and say “I like that,” or “Hey moron this would better” or anything at all. It’s not that we can’t solve problems by ourselves, it’s that we don’t want to.

I am sitting here in the Westin Taipei watching taxis among scooters among foreign signs and talking to you.

Join us.

www.opencompute.org – proudly hosted by Rackspace

For more on the Open Compute Summit, heck out this video montage capturing A Day In The Life at the Open Compute Summit 2012.

As a principal engineer with Rackspace I have had the good fortune to be involved with these efforts since the launch. Prior to our involvement with OCP, I had been part of a team tasked with development of our next generation data center strategy. Our prior pursuits had been tinged with the desire to develop infrastructure designs built upon distinctive intellectual property providing the basis for a competitively differentiated solution.

From the inception of the project, our leaders recognized the philosophical synergy between Open Compute and our OpenStack pursuits and willingly embraced the open source attributes of OCP. Further, they recognized that the collaboration and innovative spirit of Open Compute had the potential to elicit positive outcomes for a much wider audience. I was quickly swept off into activities geared to shape Open Compute designs for Rackspace consumption. Further, the open and collaborative nature of the project presented the opportunity to redefine Rackspace infrastructure requirements in lockstep with our customers, peers and even industry competitors.

With the next design summit (May 2 through May 3 at Rackspace’s San Antonio headquarters) rapidly approaching I want to take a few moments to reflect on some of the perspectives and tenets that form the “why of Open Compute” from a contributor and consumer perspective. 

Collaboration.  ‘Offset and magnify each others talents’

Here at Rackspace we place considerable value on the inherent capabilities of an individual. Internally, roles and opportunities are aligned with the strengths of the “Racker” (Rackspace employee) and tools such as StrengthsFinder are used to guide individuals in opportunistic use of their talents. Likewise, Open Compute sets the scene for such behavior on a grander community or multi-organizational scale.

Broad collaborative pursuits in data center design and implementation can help to insure that innovative principles are not confined in unnecessarily differentiated competitive bubbles. Teams can focus on areas of key expertise by building upon the contributions of others, minimizing unnecessary iteration in areas where problems have already been solved.

I have been amazed at the openness and excitement with which requirement and design discussions are embraced by this community. Even as I write this, new avenues of partnership and contribution are formulating. This does not mean, however, that the seats are full.  We are just getting started. We need and welcome participation and perspective.

Efficiency.  ‘Design philosophy, vanity free – well beyond PUE’

Amir Michael, one of the founding architects of Open Compute described the platform as “Vanity Free.” I found this description and design philosophy particularly appealing. The reductive, function first approach illustrated throughout the OCP project helps to engender a keen focus on value.

To me, the mantra “Vanity Free” is in many ways synonymous with “Value First.”  I see this philosophy also representative of the overall focus on efficiency. In the data center world, we often equate efficiency with metrics such as reduced power usage effectiveness (PUE) or optimized power supply output. Appropriate focus on such critical metrics is paramount, and Open Compute demonstrates leadership in both areas.

OCP and Vanity Free design, however, opened up a new and broader focus arena for me where attributes ranging from raw material selection and manufacturing yield complement considerations around assembly activities and shipping weight. The application of simple, functionally aligned and ground up efficiency principles in a holistic fashion is already helping my team uncover new areas of opportunity and value.

Innovation.  ‘New opportunities for real innovators’

Some have asserted that the open source nature of OCP platform design pursuits will culminate in a “race to the bottom.” In this argument, it is feared that any distinctive contribution will be quickly commoditized across a diverse fulfillment pipeline compromising the opportunity for those with innovative contributions to profit reasonably from their efforts.

Alternatively, it is my belief that OCP opens up new opportunities for real innovators to distinguish themselves and their contributions. Yael Maguire, John Kenevey, Frank Frankovsky and others have spent considerable effort rationalizing the licensing framework and have produced a model that provides what I view to be a vehicle for deep innovation into the very building blocks that comprise the resources we will consume. It is true that our lens will be keenly focused on value-oriented, distinctive differentiation. Ideally this focus will help us put a stop to what I often term the revolving door of “me too innovation” or in Andy Bechtolsheim’s terms “gratuitous differentiation” around our products.

Here, aligned design principles (simplicity/vanity free/value driven) mesh with the collaborative nature of the community to set the scene for an environment where innovators may both profit and thrive. For this to take shape, we need activity often and early from contributors and consumers alike. This involvement will help to insure that community requirements and road maps are well rationalized, that innovators are plugging in to solve real problems and that design frameworks are formulated to incorporate distinctive modules where sensible.