DEV Community

loading...
Sesame by ITRenew Engineering

Circular Data Center, Open Hardware and Open Compute Project

sramkrishna profile image Sriram Ramkrishna ・5 min read

I hope everyone enjoyed the last post about open system firmware and the role that the circular data center plays in it.
Previous post: Circular data center - Working with Ecosystems.

One of the most important umbrella projects that the circular data center works under, and one of main enablers of the circular data center, is the Open Compute Project (OCP).

Open Compute Project History

OCP was initially started by Facebook as a way to share designs of modular data center components with the rest of the industry back in 2011.

The data center industry is dominated by a number of industrial giants each with their own proprietary set of technologies, built on a set of common component building blocks. Often, the same set of technology is implemented with a specific set of value adds or design constraints. The open compute project works by convincing the industry giants, and many of the key component providers, to work together by collaborating on common technologies without the overhead of maintaining many discrete designs or repetitive bilateral relationships. By working together we can reduce costs and duplicated efforts, while also building better technology.

OCP is separated into a number of projects. I'll focus on a couple important ones that we care about, because as a vendor that sells hyperscale technology they are integral to how our products are built.

Server

The server project is one of the more influential projects for the circular data center. This project provides the physical design for the racks and chassis documenting the specifications for electrical, mechanical, and debug/test framework.

All our products are based on the open rack compatible chassis and sleds. There is a difference between the open rack vs the standard rack sizes in which rack sizes are a little taller than standard sizes - so an important ecosystem to keep track of. But the advent of this project allows the circular data center to depend on the community of companies whose expertise is building rack hardware.

That doesn't mean we are sitting idle - our contributions in this space have been around re-envisioning OCP hardware in different form factors. The Sesame by ITRenew Fast-Start system for instance is an excellent example of rack scale form factor re-imagined into a desk side form factor, and we continue with form factors for edge. In this way we take advantage of the work done by the community and create new products and new uses for re-certified hardware.

Hardware Management

This is a relatively new group - which is why it is in incubation - but it focuses on hardware manageability. Examples would be a rack manager, the BMC (baseboard management controller), telemetry and so on.

This group is an important component of data center management and is an obvious choice for circular data center companies to be involved in.

Networking

There are a few interesting hardware and software networking open compute projects.

In the software space, some companies are collaborating with each other to create community supported networking operating systems. For now, the project to watch out for is SONiC/SAI. SONiC is a project incubated by Microsoft for Azure. This is a complete network operating system that is based on Linux that runs on switch hardware from many participating vendors and is already production worthy. For the circular data center, this means with the switches we get, we can deploy SONiC to get features like dynamic breakout on recertified hardware. Projects like SONiC allow us to not worry about end-of-life software on re-certified equipment. That shows the power of community!

SAI is the interface API that SONiC implements. This opens the possibilities of writing interesting software that interacts with the switch software. With the freedom to build, the SONiC community will have innovative solutions in the near future that I believe will outpace proprietary solutions.

Rack and Power

Another important project under the Open Compute Project is the Rack and Power group. This group focuses on rack standards, they analyze power distribution, the heating and cooling, air flow and other things that makes a rack an effective and scalable solution to house hyperscale servers. We rely on this team to provide the incremental equipment to make the best solution for a data center that reduces the impact of the by products of data centers like heat, energy consumption and what not.

Conclusion

All our products at ITRenew are based on open rack compatible chassis and sleds. Mechanical and electrical compatibility across generations of OCP hardware designs allow us to build rack solutions with a mixture of node types, including combinations of storage and compute elements. These shared specifications and details are also critical information for us as we work on design extensions that allow us to expand server use cases and re-use cases. Without open availability of detailed design files and specifications, the ideal of a circular data center would be much more difficult to achieve.

Furthermore, the software like the network operating system SONiC, allows us to run them on re-certified switches allowing them a new life in the data center.

Hyperscale size service providers often retire their data center IT equipment after 3 or 4 years of service. The application demands and business growth typically drive the need for the most advanced hardware. This hardware is very useful for many businesses and applications and repurposing the hardware earns non-trivial contribution towards the reduction of greenhouse gases (GHG). Today, a large portion of the total GHG emissions come during the pre-use phase, or manufacturing phase when raw materials are mined from the earth and converted into IT equipment. In fact, when IT equipment is installed into a data center that only uses renewable energy, all the GHG released into the atmosphere is during the pre-use phase.

How does all this relate to the Open Compute Project? OCP has embraced the theory of a circular economy (CE) where IT equipment is repurposed rather than recycled. Repurposing may occur within a company or may go through refurbishment and deployed by a 2nd or even 3rd tier business. To facilitate the repurposing, many of the OCP projects are taking steps to make it easier to repurpose the IT equipment and for the new owner to achieve ongoing business continuity (firmware refreshes, security features, etc.) OCP offers a marketplace to help businesses find efficient and scalable CE equipment. Companies like ITRenew is also a critical link in this supply chain.

So to conclude this blog post the OCP Project and its members recognize the importance of the circular economy and its far reaching impact both for its positive impact on carbon footprint but also as an practical demonstration of the benefits of OCP as touched on in previous blog posts. In turn, the circular data center spends a lot of time providing OCP solutions into markets that would not normally see this, thanks to the myriad of solutions that target second and third tier markets.

That's all for this month. Next month - I will write about the openBMC project,and maybe some of the projects we are working on that y'alls might find interesting.

This blog post by reviewed by the kind folks at Open Compute Project staff, and ITRenew comrades.

Discussion (0)

pic
Editor guide