The Open Compute Project is revolutionizing the data center industry. It was created in 2011 by Facebook, and supported by Apple, Google, Microsoft, IBM, Cisco, Juniper, Nokia, among other hundreds of companies, to develop solutions for achieving hyperscalability in data centers.
OCP specifications provide the following concrete benefits:
- Energy Efficiency (lower OPEX): Technical innovations in the design of the entire infrastructure has allowed Facebook to lead the ranking with the lowest PUE in the world.
- High Density (lower CAPEX): A design that maximizes computing power per unit of space.
- Space Availability (lower CAPEX/OPEX): A new rack, with power bar, allows for fewer cables, and a cleaner installation.
- Serviceability (lower OPEX): Everything is designed to be fixed and replaced simply, without tools, and hot-swappable.
- Longer life cycles (ROI): An OCP installation evolves over many years, modularly updating the various components.
New form factors
The OCP specifications are available for any manufacturer to produce OCP-compliant products, which includes racks, power-supplies, servers, cards, among other components.
Server solutions propose a new form factor that allows for more compute capacity per rack unit:
Storage solutions are optimized for either mechanic or all-flash disks, providing the biggest storage capacity per rack unit:
Power efficiency
The biggest data centers in the world have obtained minimal PUE values (where PUE is “Total Facility Power” divided by “IT Equipment Power”), in the range of 1.06 to 1.11. Many factors contribute to this innovation, including:
Most efficient power supplies:
Centralized power supplies, with a single power tray that energizes the whole rack:
New rack unit of 48mm (OpenU), allowing for 96mm tall servers (2 OUs), which are able to have bigger fans and taller dissipators:
Better thermal design, related to the location of each component to optimize the air flow:
Visit our Open Compute Products page for more details!