Interesting problem how to organize information in Facebook's Open Compute Project

I got a chance to meet some of the Facebook's Open Compute Project team last week, and the meeting went much better than I expected.  One of the great questions Facebook asked was how to organize the Open Compute Project's efforts.  One typical approach would be a taxonomy of the different parts of the system - power, cooling, servers, etc.

taxonomic scheme, is a particular classification ("the taxonomy of ..."), arranged in a hierarchical structure.

A hierarchical approach makes sense for a technical crowd

A hierarchy (Greek: hierarchia (ἱεραρχία), from hierarches, "leader of sacred rites") is an arrangement of items (objects, names, values, categories, etc.) in which the items are represented as being "above," "below," or "at the same level as" one another.

A different way to look at the problem is to use an ontological approach and use knowledge management techniques.

DEFINITION

In the context of computer and information sciences, an ontology defines a set of representational primitives with which to model a domain of knowledge or discourse. The representational primitives are typically classes (or sets), attributes (or properties), and relationships (or relations among class members). The definitions of the representational primitives include information about their meaning and constraints on their logically consistent application. In the context of database systems, ontology can be viewed as a level of abstraction of data models, analogous to hierarchical and relational models, but intended for modeling knowledge about individuals, their attributes, and their relationships to other individuals.

Note the yellow text where ontological vs. hierarchical is compared. 

I think the ontological approach could work for Open Compute Project.  I'll spend more time over the next couple of weeks circulating the idea and getting feedback.

 

5 points amongst many others to think about if you buy from an ODM server vs. OEM

With Facebook sharing its server design and use of Quanta for its server supply.

Join Us

We worked with Alfa Tech, AMD, Delta, Intel,Power-One, Quanta and Synnex to develop of the first generation of technologies.

There are now bunches of IT purchasing departments looking to save money by buying direct and cutting out the middle man.  I’ve been laughing with my friends who have managed outsourced manufacturing projects on how naïve people are to how complex and difficult it is to buy direct from a manufacturer.  I got most of my hardware sourcing experience taking dozens and dozens trips to Japan, Korea, Taiwan, and China while I worked for Apple Computer, managing various peripheral projects for the Macintosh.  There is a huge reason why many companies have dual suppliers to manage price and quality, keeping the supplier constantly competing and knowing they will lose business if they can’t keep up.

To give you an idea of what few will hear about the Quanta Servers going to Facebook, check out this rumor of Quanta getting 20% margin on servers when their typical OEM server margin is 4%.

Order by the Facebook server Quanta: gross margin over 20%

April 18, 2011

Beijing the morning of April 18, according to informed sources, Taiwan’s contract manufacturers Quanta Facebook server to obtain orders, gross profit margin as high as 20% or more.

Reportedly, these servers will start shipping this fall, the order is Quanta gross margin is 5 times the daily orders, up to 20%.

Quanta declined to comment on.

Facebook server orders usually get HP and Dell. The Quanta favor, marking the influence of Taiwan’s contract manufacturers is growing.

At Apple we had teams who knew component cost, quality and performance issues.  They managed Bill of Materials (BOM) had HW engineering, Quality engineering, and purchasing expertise.  This is a big part of Apple’s IP that rarely gets out, and they’ve been working on outsourced manufacturing for decades.  Who would leave Apple’s HW team to go work on Amazon Kindle, HP, Barnes & Noble, etc.   I worked in the Apple purchasing group 25 years ago, and it was a blast in the Mac days. Smile

Even the Wal-mart purchasing staff don’t know how to do things at Apple’s level.  Have you ever seen a retail hit coming with a Wal-mart brand that was designed and developed by Wal-mart? 

Dell has attempted to educate some people with 5 points to consider in when choosing a server vendor.

Whitepaper: 5 points to consider when choosing a Server Vendor for Hyperscale Data Centers

Rate This

A whitepaper came out a little while ago from the management consulting firm, PRTM, that gives a perspective on the server industry.  The paper, which Dell was one of the contributors to, specifically focuses on something near and dear to our hearts, hyperscale data centers.

The paper, entitled Hyperscale Data Centers: Value of a Server Brand, talks about what organizations who are looking to build out these ginormous data centers should consider when selecting a system vendor.

But, there will be many who will dismiss this advice coming from a server OEM.

Want to know if I am right?  Watch the Server ODMs profit margins increase as many more people buy direct.

What most people don’t know is the ODMs are very clever in making it appear like they are supplying high quality just as good as the OEMs.  Cut out the middle man drop the price a bit.  But, there are many ways to cut costs which gives the ODM a 20% margin vs. 4% going through HP and Dell.  So, you think you have the game figured out and want a lower price.  And, the ODM looks for more ways to cut costs. 

Can you outsmart a huge team of Chinese engineers and business people with your one, two, maybe three people spec’ing a server?

Do you Trust your server supplier?

We’ve all seen disastrous product recalls on technology products.  Imagine 100s or 1,000s of servers being recalled.

For example, the unsexy capacitor has caused millions of dollars of damage.

Faulty capacitors have been discovered in motherboards as old as Socket 7 and have affected equipment manufactured up to at least 2007. The motherboard companies assembled and sold boards with faulty caps sourced from other manufacturers (see below). Major vendors such as Intel, Dell and HP were affected.[2] Circa 2005 Dell spent some US $150 million replacing motherboards entirely and another $150 million on the logistics of determining whether a system is in need of replacement. HP reportedly purged its product line in 2004. The motherboards and power supplies in Apple iMac G5s[3] and some eMacs[4] were also affected.

Welcome to the world of managing a ODM.  Late night regular phone calls to Asia, last minute war room meetings to address quality or supply issues.  Making the decision after weeks of phone calls, faxe’s, and e-mails, you need to get on a plane to China to resolve the issues.

Oh one other thing, at Apple it was worthwhile to do all this because the volumes were in 100,000s per month.  Only the server OEMs buy in this quantity.  As soon as you drop to 10,000 per month or less, you lose a lot of leverage with the ODM and you don’t have the benefit of large volumes to spread out your support costs.

50 Open Source Machines, imagine Facebook’s Open Compute Project doing the same

TED conference has a presentation by Marcin Jakubowski on Open-sourced blueprints for civilization.  Here is a video of Marcin’s talk.  It is 4:11.

Here are excerpts from the transcript.

We've identified the 50 most important machines that we think it takes for modern life to exist -- things from tractors, bread ovens, circuit makers. Then we set out to create an open source, DIY, do it yourself version that anyone can build and maintain at a fraction of the cost. We call this the Global Village Construction Set.

Here is a radical idea. Can you imagine a data center built like this?

I realized that the truly appropriate, low-cost tools that I needed to start a sustainable farm and settlement just didn't exist yet. I needed tools that were robust, modular, highly efficient and optimized, low-cost, made from local and recycled materials that would last a lifetime, not designed for obsolescence. I found that I would have to build them myself. So I did just that. And I tested them.And I found that industrial productivity can be achieved on a small scale.

Now you wouldn’t build this in the US, but maybe you could take this approach in an emerging market.

Wouldn’t it be cool if Facebook’s Open Compute Project supported the same vision of Marcin?

If this idea is truly sound, then the implications are significant. A greater distribution of the means of production, environmentally sound supply chains,and a newly-relevant DIY maker culture can hope to transcend artificial scarcity. We're exploring the limits of what we all can do to make a better world with open hardware technology.

Thank you.

(Applause)

Is Facebook the Lightning Rod for Environmental Impact of Data Centers?

DataCenterKnowledge posts on Facebook's 100KW Solar array.

Facebook Installs Solar Panels at New Data Center

April 16th, 2011 : Rich Miller

Facebook has built a large solar array next to its new data center in Prineville, Oregon (Photo: Rich Miller, Data Center Knowledge).

Facebook has installed a large array of solar panels at its new data center in Prineville, Oregon, which will supplement the local utility in providing electricity for the 300,000 square foot facility, which was officially opened at a ceremony yesterday.

For more than a year, the environmental group Greenpeace International has been bashing Facebook over its use of electricity generated by “dirty coal” to support its huge new Oregon data center  – a campaign that continued Friday as Facebook opened the new facility.

The photo Rich has of the solar array on a cloudy day reminded me of a post I wrote 2 years ago on who Greenpeace's first data center target would be.

What is the First Greenpeace Data Center Target? Apple? Google? Microsoft?

TUESDAY, JULY 28, 2009 AT 9:20AM

Datacenterknowledge blogs on how quickly Apple is building its $1 billion dollar data center.

APPLE MOVING QUICKLY ON NC PROJECT
July 28th, 2009 : Rich Miller

apple-ncApple is known for keeping its new technology secret prior to launch.

And, I closed with the following.

How can Greenpeace, not already have a plan in place to address Apple’s data center for its environmental impact?

Now, you could say Greenpeace why not go after Google or Microsoft?  Greenpeace could, but why haven’t they already.  It is not worth it for media coverage.  Going after Apple would get people’s attention.

If not Apple, who else makes sense to go after if you were Greenpeace?

With Earth Day coming up Facebook is the one company as a target for Greenpeace and the rest of the data center industry is relieved.  Facebook is a convenient target being the only data center operator I know that had an option to source hydro or coal power for its data center and selected coal.

When Greenpeace went after environmental impact of PC products they targeted HP, Dell, and Apple.  Is it time for Apple, Microsoft, and Google to feel the attention of Greenpeace or is Facebook the only high carbon impact data center user out there?

I'll be at Green:NET 2011 later this week and will be in this session.

GREENPEACE INTERNATIONAL WORKSHOP – ROOM 2

Dirty Data: How the Cloud is Powered and Why it Matters for the Climate

Greenpeace will release a new analysis that looks at leading IT companies (Facebook, Apple, Google, and more) and asks if IT, as it builds out the cloud, will perpetuate the dirty energy issues of older, entrenched industries or will be the innovative sector that creates a business model that prioritizes a future built on clean, renewable energy?

Speakers:Gary Cook - Senior IT Policy Analyst, Greenpeace International