The Open Compute Project, in which hardware manufacturers and large cloud operators jointly design data center-compatible hardware according to open source principles, has presented its strategy for the next few years. Founded ten years ago from a Facebook initiative for energy-efficient data center hardware, companies from Facebook to Microsoft, from Intel to Nvidia are working here on open specifications for servers, switches, routers, storage, power supplies and racks.
Stand for the next few years under the heading OCP 2.0 a number of topics on the list that are being pursued in several project groups. One of them is modularity: The OCP wants to specify scalable components and open interfaces in order to facilitate hardware development. Another goal is to improve the cross-platform management of OCP hardware.
A stronger standardization of metrics for utilization and energy consumption as well as OCP-certified operating modes should improve sustainability. In addition, one would like to certify vertical stacks for certain use cases.
All of these measures are intended to meet the requirements of the market and thus contribute to the further spread of OCP hardware. Currently going almost 90 percent of the OCP hardware goes to large cloud operatorswho are mostly OCP members. Less than ten percent end up in other companies’ data centers. The OCP would like to become stronger here.
On the technical side, the use of optical technology in the data center, the development of a standardized infrastructure for artificial intelligence and machine learning as well as improved cooling technology are on the agenda. With this, the OCP would like to position itself for future technical advances.