John Hu is VP of Inspur Group Co, Ltd.
Bill Carter is CTO of Open Compute Project Foundation.
There’s a lot of talk in the industry today around the evolution of openness. The industry is experiencing unprecedented increases in demand for IT resources and a continued need to deliver technology that drives efficiency, capacity, and throughput. These challenges impacting the data center are being addressed through collaboration and open sourcing in order to share the cost of developing products and technologies, and to accelerate the advancement of core technologies. In effect, idea sharing and collaboration are mechanisms driving the open source ecosystem.
For data centers in general, and even more so with hyperscale data centers, improving energy efficiency with workload optimized hardware remains a key objective. One positive impact of OCP equipment as it relates to less energy consumption is through the disaggregation of software and hardware, which generally leads to a reduction of the number of servers and overall energy consumption. What’s more, virtualization allows for workloads to move across the data center at different times with varying processing demands, helping to alleviate hotspots and improving quality in the density of each rack. For hyperscale data centers, OCP members are involved in helping to optimize racks to better handle the intensive, data-driven applications such as machine learning and artificial intelligence (AI). One area where open sourcing can play a key role is in sharing these designs and best practices with a broader customer base, helping to promote more efficient design and use of servers to handle specific and arduous processing workloads.
Collaboration, Innovation and Competition
Vendors committed to openness tend to be active in any of the open source communities, such as OCP, ODCC and Open19. What’s become clear is that there is a lot of opportunity to encourage communication across this entire open ecosystem and to work together to make openness collaborative rather than continuing to work toward an older model of keeping proprietary vendor offerings secret. Open sourcing has already demonstrated that it helps foster collaboration and drives innovation in software, but we’re now extending that model to hardware and embedded software. The latest OCP Standard Rack Server realized the integration of OpenBMC technology and Redfish standard.
An open community is committed to creating open hardware solutions working with data centers and their customers, and sharing lessons learned broadly so that the industry as a whole can realize the benefits of pursuing openness in the design and functionality of their equipment. Competition still exists, but is based more on supply chain management and a vendor’s ability to respond to industry demand and emerging technologies that bring platforms to market more quickly and efficiently. In doing this, there is a general consensus that collaboration and innovation will continue to expand, and data centers will be able to meet growing service demands from new technologies being adopted by enterprises so that everyone benefits.
Embracing options is essential in an open community and OCP is a foundation that fully embraces the idea that there are a lot of different types of customers with different needs and they should have an opportunity to choose the best architectural that meets their individual requirements. For example, while OCP supports a wide range of open-sourced rack level 19-inch EIA compatible products, the organization also accepted and supports the Microsoft contributed Olympus architecture—a rack architecture that fits a standard EIA-310 19-inch rack and provides an efficient, phase-balanced power delivery scheme. Because OCP serves a global community with broad requirements, there’s an understanding that no two data centers have the same set of requirements. OCP works with its community to provide open source options that meet the global demands for cloud hardware. Today, that is being accomplished with three similar yet different architectures.
A recognized opportunity and a technology ripe for convergence is management of IT resources. That management occurs with each device but can also be handled as an aggregated solution across the rack. This rack level function can be performed with a low-cost Rack Management Controller (RMC). Today, we see a number of suppliers and solution providers already collaborating on Redfish management standard. The open source community is also discussing ways to collaborate on open source rack level technology that will support different physical architectures. This is a clear example how fostering openness leads to shared ideas that generate broader interest and participation that can not only improve product offerings but streamline their development, decrease costs and lessen the time to market.
To effectively extend cloud technology and products to enterprise customers, suppliers who often view themselves as competitors need to take a different approach to deliver viable solutions. The modular, scaleable design of OCP’s openRack or Olympus, along with the functionality of a management system based on open source, ensures a balance between rack performance and efficient power consumption, effectively lowering OPEX and CAPEX and providing data centers future-proof hardware for growth and scale out over the long term.
Opinions expressed in the article above do not necessarily reflect the opinions of Data Center Knowledge and Informa.