Robert Bunger, director of Data Centre Industry Alliances, Schneider Electric, provides an insight into how data centre operators can adopt Open Compute Project principles
The Open Compute Project (OCP) was born out of the necessity for internet giants to quickly assemble massive hyperscale data centres to support their business models cost-effectively. Whereas IT vendors, at the time, had product lines created to address a diverse customer base ranging from home and small business customers up to the largest corporate or enterprise clients and big data centres, the hyperscalers such as Facebook, Google and Microsoft saw the need for IT products that were optimised for the massive cloud data centres they were creating to deliver services.
Seeing the potential to cut out unnecessary costs from their businesses, they sought to develop their own designs, which could be produced in high volume and with great economies of scale, using unbranded equipment supplied by original design manufacturers (ODMs).
The goal of the OCP is to “enable delivery of the delivery of the most efficient designs for scalable computing”. The foundation believes that openly sharing ideas, specifications and other intellectual property “is the key to maximising innovation and reducing complexity”, and that “sharing intellectual property with others helps the IT industry to evolve” and to “throw off the shackles of proprietary ‘one size fits all’ gear”.
Challenges to adopting the OCP approach?
To date, the widespread adoption by the industry of OCP hardware has not been as strong as might have been hoped. Many of those who have chosen to deploy OCP-compliant solutions are large end users that initiated the foundation and remain the main contributors of ideas and designs for hardware. As such, the supply chain reflects the needs of these very large customers. Many of the ODMs building products to OCP specifications are focused on producing large volumes and are less inclined to build designs in lower volumes for niche customers.
Also, there is a cost overhead associated with installing OCP-compliant hardware. It is not as easy to ‘plug and play’ OCP equipment as it is with converged architectures or standardised IT gear. Specialist expertise is needed to establish an OCP data centre but the payoff is that, once it is operational, it is easier to maintain and requires fewer people to manage.
Addressing issues such as these requires the emergence of an ecosystem of product vendors, systems integrators and maintenance service providers willing to address other areas of the IT market. Specifically, those other than the largest data centre operators. There is some evidence that this is beginning to happen with, for example, one system integrator in France, a Platinum OCP member, building systems aimed at the gaming industry. Other integrators are building and supporting systems in other niche markets, thereby broadening the appeal of the OCP.
Essentially, what any adopter of OCP systems needs to know is who will build the products, who can integrate them into a data centre to meet my requirements, and who will provide ongoing support?
How do brands get involved with OCP?
As the original impetus for the OCP came from large organisations deploying equipment that was essentially custom built to their own specifications, such products tended to be ‘vanity-free’, or unbranded by the traditional familiar IT vendors.
As the market for the OCP broadens, brands will become more important as customers seek the trust and reassurance promised by purchasing products from established vendors, that are in turn exploring the benefits to their own businesses of engaging with the foundation.
Schneider Electric, for example, was interested by some of the power-related features in OCP specifications and saw, on examination, that centralised power supplies presented a broad market opportunity for its own expertise.
Accordingly, the company developed a compliant rack and power supply solution, which were both submitted to the foundation. The power supply design meets a broad market need in terms of density and scalability and as such it is helping to broaden the appeal of OCP-compliant systems beyond the current installed base of large data centres.
What can you do to adopt OCP principles?
Membership of the foundation is not restricted to vendors; on the contrary, participation by organisations throughout the supply chain is encouraged from vendors, systems integrators through to maintenance service providers and end-users. Participation is even open to non-members; every meeting is open to the public, all are recorded and can be played back via YouTube, which is another way the OCP continues to drive awareness of its benefits to the industry.
The OCP is essentially a forum for sharing designs and expertise. The principles are straightforward: efficiency, scalability, openness and impact.
What are the main benefits of OCP?
Organisations can benefit in several different ways depending on their position in the market. Vendors can, at a minimum, gain mindshare through participating with OCP.
Whether current adoption levels remain high or low is not important; people are still watching its developments and there are many benefits in being seen to participate in driving technology forward.
Ideas submitted to OCP can spark developments in a vendor’s own product portfolio, as in the case with Schneider Electric and its development of the centralised power supply and OCP-ready rack.
For systems integrators there is the opportunity to be part of a new eco system, which supports an emerging market for standards-based solutions designed and built by collaboration between key players in the industry. Therefore, the OCP enables both vendors and end-users to join as members, which can submit product designs and open reference architectures that can be used to form the basis for new critical infrastructure applications.
From a customer’s point of view, the benefit of using OCP equipment is efficiency both in terms of energy and operations. OCP technology delivers more compute power per kW, helping to reduce costs and meet higher environmental and energy efficiency standards, such as an improved PUE rating. As hardware is optimised for the task in hand, investments in equipment costs are reduced and with a focus on scalability, staff numbers do not have to increase as the volume of equipment in a facility does.