Project

OCP Strategic Initiatives are higher-level groups for collaboration across and between the OCP Projects, particularly to solve the co-design required for AI infrastructure. Three initiatives are active. Open Systems for AI is the umbrella that coordinates the Community's work across Projects and Initiatives, with its own steering committee to identify goals and gaps. Open Data Centers for AI tackles the challenges of delivering massive AI data centers at scale, from order-of-magnitude increases in rack power density to GW-scale facilities and multi-GW campuses. Open Cluster Designs for AI develops Open Pod Group and Open Cluster designs, with guidelines spanning small, enterprise, neocloud, and hyperscale deployments.

Incubation Initiatives is a group of early-stage OCP initiatives: community efforts being incubated with the goal of graduating into a full OCP Project or Sub-Project. The group currently includes the Digital Twin Initiative, which explores how digital twins can enhance the design, deployment, operation, and optimization of open hardware in data centers, and the Test and Validation (T&V) Enablement Initiative, which works to open-source the software that tests open hardware. Longer-horizon technology work is incubated separately by the Future Technologies Initiative. Details about the incubation process and ways to get involved will be added in the near future.

The OCP Networking Project is creating disaggregated, fully open networking technologies that allow rapid innovation in the network space. It facilitates the development of network hardware and software, together with trusted project validation and testing, in an open, collaborative community, bringing to networking the principles OCP brought to servers and storage so end users can move beyond closed, proprietary switches. Its initial goal is a top-of-rack (leaf) switch, with spine switches and other hardware and software to follow. Scope covers open switch hardware, Linux-based operating systems, automated provisioning, integration, efficient power and cooling, and SDN.

The Cooling Environments Project is the OCP community focused on advanced data center cooling. High performance computing has long used liquid cooling as a cost-effective, efficient way to extract heat. As 5G, IoT, VR, CDN, and other latency-sensitive applications push data centers into regions where cooling is difficult, and as power densities rise, liquid cooling, specifically warm water cooling, becomes an effective alternative. The project focuses on five functional areas of the data center: cold plate, coolant distribution unit, immersion, door heat exchanger, and heat reuse.

The Time Appliances Project (TAP) addresses time synchronization, a key element of efficiency in distributed systems. Industries from telecom and mobile to power, industrial and professional audio/video all need accurate, reliable distribution of time and frequency across packet networks. TAP provides a platform to bring together, discuss, standardize and share timing technologies, with datacenter applications and network infrastructure as its main interest, uniting datacenter operators, application developers, and equipment and semiconductor companies to enable time-sensitive applications such as distributed consistency, edge computing, AR/VR and IoT. The project supports a datacenter PTP profile covering PTP-aware switches, clocks, NICs and timing modules, and promotes open-source PTP implementations.

The OCP Storage Project focuses on chassis and sleds, components and peripherals, network-enabled storage, and compatibility solutions. It furthers the goals of standardization and modularization by channeling investigations into three primary areas: storage devices, storage-related chassis, and storage technologies.

The OCP Server Project provides standardized server system specifications for scale computing. Standardization keeps the OCP specification pool from fragmenting into point solutions, and the project works with the other OCP disciplines to ensure broad adoption and optimization across validation, manufacturing, deployment, data center operations, and de-commissioning. Its scope covers Open Rack and Open CloudServer compatible chassis, sleds, and blades; micro-servers; EIA 310-D/E 19-inch rack derivative chassis and sleds; components and peripherals such as riser cards, mezzanine I/O, accelerators, and networking; and specification standards for interfaces, manageability, debug, and test.

The Rack & Power Project Group is the first group to focus on rack standards designed for data centers, integrating the rack into data center infrastructure as part of OCP’s “grid to gates” philosophy, a holistic design process that considers the interdependence of everything from the power grid to the gates in each chip. Its scope covers specs, standards, and products supporting rack system architectures (19-inch EIA compatible and 21-inch Open Rack), rack-level power systems for conversion, storage, and control of power between facility infrastructure and IT gear, and the shelves, supports, sub-chassis, and adapters needed to implement IT gear.

Open Platform Firmware (OPF) is an open development project whose goal is to let system owners "own their firmware" by moving the point of control to the owner, who must be able to change firmware and share it with other owners. Since March 2021, OCP server badging has required OPF support. OPF is open source, but vendors may deliver compliant firmware either fully open, as with IBM's Power 9 systems, or as componentized binary images such as UEFI into which open source components can be integrated. LinuxBoot and OpenEDK2 are examples. Its scope spans all data center processor architectures, cloud operating systems, compute, storage and network devices, tooling, and security.