The Open Compute Project Foundation (OCP) celebrates its tenth anniversary this year and continues with its mission to apply the benefits of open source and open collaboration to hardware and rapidly increase the pace of innovation in, near and around the data centre. The OCP is a collaborative community focused on redesigning hardware technology to efficiently support the growing demands on compute infrastructure.
An integral part of this mission is to foster innovation and to facilitate this aim the OCP Future Technologies Initiative (FTI), a formal project within the Foundation, whose mission is to build a future-focused community within OCP to serve as forward-looking funnel for ideas and technologies three-to-five years into the future was created.
Two years ago, the Foundation initiated the OCP Future Technologies Symposium, an annual event that brings together academia, the technology industry, start-ups, venture capital and analyst firms to solve future challenges facing the industry, and accelerate productisation through partnerships and open source R&D. All submissions are featured, and the winners are announced at the Summit events OCP holds each year.
The OCP Future Technologies Symposium effort has been successfully led by Chairs Allan Smith, currently the Lab Manager for Area 404, Facebook’s hardware prototyping laboratory, and Lesya Dymyd, a strategic innovation engineer at 2CRSi, serving as the Symposium Program Chair. The next OCP Future Technologies Symposium will be held on November 8 in San Jose, CA, in conjunction with the 2021 OCP Global Summit.
“My background is from NASA, and I played a role in developing the small satellite market and commercialising space assets,” Allan Smith, OCP Future Technologies Initiative (FTI) Chair says. “Any time we had a gathering in those worlds it would include academic outreach element. When I joined this industry in 2016 and attended the first couple of OCP events, I realised that this was missing and that is the role that I play in hardware research. I went to the leadership of OCP and said, ‘why don’t you have this symposium element?’ They agreed it was a great idea and told me to go build it. That was January of 2019, and they wanted to have a symposium right along with our OCP summit in March of 2019.”
The Future Technologies Initiative and how OCP use future technologies are to look out to a three to five year plus horizon. The focus is on research-oriented projects and the pathway to bring both those opportunities and potential solutions to the problems that the industry is facing. Anything that is in a shorter time frame is presumed would already be under development within the industry.
“Any venture in the research or innovation realm typically says this is not something we are doing today,” Smith continues. “Whether that is in the problem space, like the reason those problems are manifesting is we do not have solutions applied to those, or in the opportunity space, we never thought about doing it this way. If you think of any start-up in the world, any researcher in the world, the goal that we can give them is actual problems to work on. Our aim is how do we give them validated use cases that are working in the industry today, whether it’s Google, Facebook, Microsoft, Alibaba, or one of the smaller developers?”
For the 2021 Symposium, the focus is on four workflows in – sustainability and data centre efficiency, power delivery and efficiency, high performance and hardware acceleration, and networking. But moving forwards, the plan is to tackle specific challenges that the sector faces. “We are taking some big, hairy, ambiguous spaces that the industry is committed to partner on and recognise they cannot solve these problems independently, or if they do, they will just spend a ton of time and resources to go solve them,” Smith adds. “We are going to spend 2021 figuring out one or two collaboration areas with very specific workloads and use cases attached to them, to provide them to this research community and say, ‘if you solve these problems, these are the biggest things we are facing – not just one company; multiple companies are facing these problems.”
For Smith the key performance indicator over the next two years is adoption of the technologies developed. He views adoption in three ways. The first measure is industry adopting and co-opting research. “This could be investing dollars into people that have talent,” he explains. “If they have some ideas that are associated with these areas that are aligned with the outcomes they are after and fund the research to accelerate those solutions.
“The second criteria is talent; wanting to hire the best talent into our workforces. The third is just adopting the solutions into our operational fleets. That is publicly what we spend most of our energy on; we filter all our decisions on how we bring those Ideas, talents, and solutions to OCP problems. The next way they can benefit if they’re joining us through FTI, is we give a small grant of $10,000 for the best paper.”
One of the main benefits of the symposium is that it delivers a pipeline of ideas flowing to the open compute project industry members through future technologies. “One language that I have learned throughout my career is what problem are we solving?” Smith adds. “When we centre on that, we will find many times that really smart engineers or researchers will get centred on a solution, like what if we could use augmented reality? Which is great, but who needs that? Or what opportunity are we chasing? And how is that going to improve our world here?
“Bringing people back to that reality becomes a very grounding discussion, and I find we are more intentional in our desire to work towards outcomes when we take that time. I am using language lately to say it is good to spend 90 per cent of our time and energy agreeing on the problem because we will work more efficiently with our last ten per cent. The biggest cultural adjustment we go through is that academic researchers and engineers are typically the kings of their domain.”
Route to market
There has been a lot written about the difficulty in commercialising technology and the so called valley of death between research and innovation, particularly in the case of new technologies developed at universities. One of the realities of life is that you cannot force things.
According to Smith this is best captured by Gordon Moore’s Law. “We are just reaching a limitation of physics,” he explains. “If you take any company in the world that deals with internet content, you are not only cataloguing the entire history of your platform, which continues to grow year over year, but every device that accesses these platforms has more cameras, more pixels, demands more bandwidth.
“The demand is growing on what our systems need to do, yet our supplier base gains are going from double every couple of years to maybe ten per cent after three years. That is an emergency brake pulled on progress and growth that we used to rely on for about 60 years in this industry, and now we no longer have that. It is taking a while for our industry to wrap our heads around the pain that we are experiencing.
It’s also one of the unique challenges throughout history because the world effectively operates online and relies on our ability to solve these problems. Therefore, it is such an exciting space to be in because there must be a transformative change here, or we are going to fundamentally fail in our contract or agreement with our customer base as an industry. If you can continue evolving how you use these platforms and services, you can continue to explore these things and your business and your personal life and add more and more content. We are going to try protecting you and make sure you are not bullied and harassed and have a fake type of information.
“When we start to reach some tough limitations of what our capabilities are to meet that growing demand, we are going to have to make some tough choices in our business models across this industry. Those are the biggest challenges and forces facing where we are at this moment. I am convinced they are going to write history books about this time and the people who stepped up to meet these challenges.
“I’m going to abstract any organisation in the world that has online content. Any company in the world is going to have a business model based on revenue, and to protect that revenue, they are going to need to offer these essential services of protecting user’s data and protecting the user’s integrity of operating on that platform.
“When we reach a limitation of what we can offer to these two sides, companies are going to be faced with very difficult choices of only having a certain amount of bandwidth, storage and processing power. Imagine a scenario where there are 120 national elections that happen in a year. Even if your interaction with social media platforms is limited to work-based platform like LinkedIn, you are still going to see this content popping up on your platform. In super highly politicised times, people are coming to your platform for a purpose, and now we are saying the problems are increasing, and we are decreasing the capability we can give to that.
“It is going to force some very hard business decisions in the coming years. Unless, and this is where the opportunity of OCP Future Technologies Initiative lies, we figure out how to solve these technical problems with big breakthroughs of new computing architectures and functions.”
The role of sustainability
Sustainability is and will continue to be one of the top priorities that both the OCP and the wide computing community face. According to Smith, the interesting thing is all the work they have put in over the past decade leads to this point where the industry now has the opportunity. “We have gathered people’s interest, and we have figured out how to align on these projects. Now we can position to solve massive multi-generational problems, like a limitation of resources.
“When we used to think we were going to use this body to solve which product, or what’s the next generation acceleration of processing power problem, we are finding that having the agreement and alignment, means and mechanisms for us to collaborate, we are now bringing that to these massive multi-generational problems. It is very different from one company just setting out and saying they will be net-zero by a certain time.
“For example, I am Company A, and I buy X number of millions of units of motherboards per year. I can say I am going to develop a philosophy with my X number of motherboards to try to figure out how to extend the life of those and to reuse those in our ecosystem. The companies that are manufacturing these motherboards are still going to have a business to run, and they are still going to manufacture motherboards. All we are doing is kicking the problem to someone else. When the entire industry aligns and says, this is going to be our philosophy to work together towards this joint goal. Now we force the suppliers to adapt to us because they do not have anyone else to sell to. That is the purpose of OCP.”