Facebook keeps score of Serviceability and Operational Efficiency of Data Center Hardware

There is a short post on OCP by Charlie Manese, Facebook Hardware Design team on Serviceability and operational efficiency, so I will just put the whole thing up.

Know the guys at Google have this data, wonder who else does?

Facebook's perspective on serviceability and operational efficiency

Wednesday, October 09, 2013 · Posted by  at 8:09 AM

UPDATED - Webinar on October 24, 2013

By Charlie Manese, Facebook Hardware Design team

At Facebook, because of our scale, we require that solutions deployed in our data center be engineered for maximum operational efficiency and serviceability.

The data center team works closely with the hardware design team to ensure this. Our designs incorporate features such as front-of-rack serviceability, toolless repair operations, and simplicity.

We’ve completed time-in-motion studies, streamlined processes for inventory and repair, and have developed scorecards to that help us evaluate and compare different hardware solutions.

Below is a table of the time-to-repair comparison of different kinds of web servers that have been deployed in our environment:

 NewImage

 

If you're interested in learning more about how Facebook thinks about serviceability and operational efficiency, and you missed the original event, I'll be joining a Hyve webinar on October 24, 2013.

 

For more information on the event, please see  https://synnex.ilinc.com/perl/ilinc/lms/register.pl?activity_id=zvkkfkw&user_id=

 

Hope to see you there!

 

Open Network Switch is closer to ship, news from Open Compute Project

Open Compute Project’s president Frank Frankovsky posted on the progress of the Open Network Switch.

Progress in opening up the network

Monday, November 11, 2013 · Posted by  at 11:20 AM

Just six months ago, we announced our intention to expand the charter of the Open Compute Project to include networking hardware.

At the time, it was our hope that we could build on the momentum we'd established in opening up server, storage, and datacenter designs and collaborate with the broader community on the development of an open, OS-agnostic top-of-rack switch. Such a switch, we believed, would enable a faster pace of innovation in the development of networking hardware; help software-defined networking continue to evolve and flourish; and ultimately provide consumers of these technologies with the freedom they need to build infrastructures that are flexible, scalable, and efficient across the entire stack.

Our progress so far has exceeded even our lofty expectations -- hundreds of people are actively collaborating on the development of more than 30 potential contributions, covering most of the network hardware stack and even some of the network software stack.

 Arstechnica adds in the threat this has to Cisco.

Six months ago, Facebook announced that its Open Compute Project (OCP) would develop a top-of-rack switch that could boot nearly any type of networking software. With the help of Intel, Broadcom, and others, the consortium devoted to open hardware specifications would develop a rival to Cisco's network hardware.

Here is the Cisco response.

In response to today's Facebook announcement, Cisco said in a statement to Ars, "It’s important to acknowledge that the largest web-scale companies driving OCP have the skills, resources, and specialized traffic patterns that justify considering this approach carefully. However, most IT departments won’t relish taking on the additional operational cost, skills and expertise that are required to integrate their own technology.

"We’re finding that the majority of customers are looking for a turnkey solution that gives them the programmability and flexibility they want, with lower operating costs, and that’s exactly what Cisco ACI will deliver, without compromises on scale, performance and security."

Open Compute Project History described in a Taxonomy Map

In Frank Frankovsky's keynote at Interop he showed the progression from red to green of the open compute project from data center, rack, server, storage and now network.

Frank started with the open source software used.

NewImage

Then Facebook shared its data center practices.

NewImage

Next came racks in the co-lo

NewImage

Eventually building to the complete taxonomy below.

NewImage

What's next?  The Open Compute Project comment is the one who asks for new things as well as Facebook deciding to contribute in an open source manner.

Open Compute Project's Frank Frankovsky launches Network Initiative

The folks at Open Compute Project are on a roll, driving open source ideas into servers, data centers, racks, etc.

NewImage

And now their latest move is networking.

With that in mind, we are today announcing a new project within OCP that will focus on developing a specification and a reference box for an open, OS-agnostic top-of-rack switch. Najam Ahmad, who runs the network engineering team at Facebook, has volunteered to lead the project, and a wide variety of organizations — including Big Switch Networks, Broadcom, Cumulus Networks, Facebook, Intel, Netronome, OpenDaylight, the Open Networking Foundation, and VMware — are already planning to participate.

Facebook is a big sponsor of Open Compute Project which is non-profit organization, and a driving force.

Who competes against Open Compute Project?  The visibility for Open Compute Project has increased and is referenced by almost all the tech media point with an article at some point.  There isn't really much competition.

What Open Compute Project does threaten is the event conference vendors.  When you have a non-profit focused on end user benefits and changing the industry, the for profit event staff who are driven to maximize revenue will maneuver to out market their event as the premier event in the industry.  Note: watch for those words when someone claims they are the premier event.  

BTW isn't it ironic that Interop has Frank keynote what a message that Open Compute Project is revolutionizing the industry.


Location: Mandalay Bay H
Wednesday, May 8, 2013, 8:30 AM-10:15 AM

Billions of people and their many devices will be coming online in the next decade, and those who are already online are living ever-more connected lives. The industry is building out a huge physical infrastructure to support this growth, but we are doing so in a largely closed fashion, inhibiting the pace of innovation and preventing us from achieving the kinds of efficiencies that might otherwise be possible.

In this keynote, Frank Frankovsky will provide an overview of the Open Compute Project, a thriving consumer-led community dedicated to addressing this issue, promoting more openness and a greater focus on scale, efficiency, and sustainability in the development of infrastructure technologies. Frank will delve into the brief history of the project and describe its vision for the future.

Seems like Interop just gave Frank a chance to market to the audience they need to go to the next Open Compute Summit and attend for free.  Or be a sponsor where there is more transparency in how their sponsorship is used.

We'll see what kind of networking guys show up to the next Open Compute Summit.  In some data center IT environments the network gear can be 50% of the IT budget.  Networks are important, but it is hard for most to accept a 50% spend.