SeaMicro uses “lean engineering” to build servers economically in Silicon Valley

My wife recently found me a great deal on Bloomberg BusinessWeek to get a 3-year subscription for $18. While reading the first issue to arrive, I found a great story about a company in Silicon Valley that has created a flexible, U.S.-based manufacturing system for their product. The company is called SeaMicro. They make low-power servers for Internet companies like Mozilla and eHarmony. I liked what the CEO, Andrew Feldman, had to say about why they have taken the unconventional approach of building their servers onshore.

Feldman says that manufacturing locally will help SeaMicro compete with bigger, deeper-pocketed rivals. The company’s engineers constantly experiment with the latest and greatest components in a bid to lower the power consumption and quicken the performance of their systems. They can then take their changes down the road to NBS—less than a mile away—and start testing them in new systems immediately. “You don’t have to deal with working across the globe and shipping stuff back and forth,” says John Turk, the vice-president of operations at SeaMicro. “You can lose days with systems sitting in Taiwan or China.”

Even more satisfying is reading how two company executives were dismayed by BusinessWeek’s line of questioning.

When asked if the happy marriage between SeaMicro and NBS will dissolve should SeaMicro hit it big and shift toward mass production, consternation fills the faces of Turk and Maslana. “It’s not about us getting big,” Turk says quickly. “It’s about how do we stay flexible. That is what the big guys don’t have.”

It was an okay article, but I have a few questions of my own for the executives. How did you get the courage to buck the trend? What are the basic principles of your operating system? Why have you chosen those principles. As an operations manager, these are three questions that I am interested in having answered.

SeaMicro: Stars and Stripes and Servers Forever

Which is more lean? Little Caesars or Papa Johns.

I have been enjoying Ron Pereira’s great blog at He recently posted the following question to readers: Is Little Caesar’s Lean?

Is Little Caesars Lean? I’d like to offer my view. A basic feature of the ideal Lean system is pull initiated one piece flow. Pull initiated means that the factory does not make the product until the customer places an order. One piece flow means that the factory is able to accommodate order sizes as low as one piece as well as high piece-to-piece feature variation with no finished goods inventory.

Pull initiated one piece flow is not how Little Caesars does things. Their key competitive advantage is cheap, instantaneous pizza service. I love those Hot-N-Ready pizzas. Any time of day, I can drive up to a Little Caesars, hop out, and pick up a couple of Cheese or Pepperoni pizzas without a second of waiting. The paradox is that Little Caesars has to make pizzas ahead of time and keep a small inventory in order to provide the instantaneous service. The other non-lean feature is that they only offer Pepperoni or Cheese. No infinite topping variation on a Hot-N-Ready pizza. You would think that Papa Johns has a better business model. At Papa Johns, the customer must place an order before the pizza is built and cooked using pre-staged dough and toppings. Their business model is pull initiated one pizza flow. In fact, that is the business model of most pizza restaurants, except Little Caesars. The typical pizza house carries no finished goods inventory, but the pull initiated one pizza flow introduces a 10-15 minute wait that prevents regular pizza houses from capitalizing on the unanticipated and emergent pizza needs of potential customers.

Although pull initiated one piece flow is supposed to be the gold standard of Lean manufacturing, it really is not. The gold standard is whatever system delivers the most value to the customer. Since a made to order pizza cannot be made in microseconds, I would rather forsake extra toppings in order to get my pizza cheap and fast. The value from my perspective is speed over variety, so Little Caesars ends up being more lean than Papa Johns because they deliver value to me like no other pizza place. Toyota recognizes this, too. As we all know, Americans like their cars delivered at the time of purchase. In order to service the great majority of American consumers, Toyota has to build and maintain thousands of cars in finished goods inventories on dealer lots. This is waste in the process that must be tolerated, because the customer requires instantaneous delivery.

Putting the cart before the horse

This morning, I experienced my annual visit to the optometrist. The wait was 30 minutes, but that is not what I wanted to address. At the optometrist, they have a fancy machine that takes a picture of your eye. I’ll call it the eyephoto. This is an expensive machine and insurance does not cover it. If you want to have the eyephoto, you have to pay $35 out of your pocket.

In order to see if you want the eyephoto, they give you a piece of paper when you walk in that has a paragraph description of the eyephoto and a place to check a box and sign if you want it done. I decided to keep my $35.

In the exam room, while waiting for the doctor, there was not much for me to do, so I was watching the screensaver on the office computer. What was it showing? It was a commercial showcasing the benefits of the eyephoto. As I watched the commercial, it occurred to me that paying the $35 might not be such a bad deal. The commercial changed my mind.

Here is an example of a badly sequenced process that does not get the customer information when the customer can use it to make a purchasing decision in your favor. It is putting the cart before the horse. I wonder what the rate of use is for the eyephoto at the optometrist. I bet it is pretty low. Wouldn’t it be higher if they reversed the sequence? If I had walked into a lobby with a giant flat screen TV showing non-stop eyephoto commercials and THEN been presented with the election form, I bet I would have signed up.

Most Lean we see addresses the movement of product before or after purchase, but in this example, an appreciation of Flow creates the potential for increased revenues. There is a huge amount of value to be gained by using Lean to ensure the purchase of product.

United Adds $25 Fee for 2nd Bag

I saw this story about United Airlines charging customers to check a SECOND bag. Not the fifth or sixth bag, but NUMBER TWO!

This reminded me of a training slide that we have in our Lean education program. There are three ways to cut costs. You can cut costs across the board by reducing all budgets a fixed percentage. This is the lazy path. You can cut costs by cutting services. This is the stupid path. Finally, you can cut waste. The smart path.

This extra fee strikes me as part of the stupid path because it cuts a core service and makes customers pay extra for something they get “free” from other airlines. According to the article, United expects it to generate $100 million in revenue and cost savings a year. Does this mean that United’s tickets will be consistently cheaper than companies that do not charge a per bag tax? I highly doubt it as the article shares that this is but one small part of a larger plan to charge more for less, a clear violation of the Profit=Price-Cost rule:

Airlines want to charge more for not only checked baggage but assigned seats and other services. Investors have urged airlines to pass on the higher costs of fuel to passengers through ticket-price increases or similar surcharges.

If United is planning to save money by flying fewer people, they might be able to claim savings because I don’t think their scheme will end up with them making any more revenue. We’re likely to see United lose revenue to the benefit of airlines that are more responsive to real flying customers, not day traders.

********UPDATE 2/26/2008**********

It appears that US Air is going to charge $25 for a second bag.


The Theory of Inventive Problem Solving

What is the Theory of Inventive Problem Solving?

The Theory of Inventive Problem Solving, also known as TRIZ, is a system of rules and tools aimed at practical problem solving. It was originally geared toward patents within the engineering community, but also applicable to many other disciplines including technology forecasting, strategic planning, etc. Basically, its an iterative process for systematic innovation that teaches you how to find answers to your problems, often by looking at other scientific fields. An underlying concept is that somebody, somewhere has already solved your problem —- the challenge is to find that solution and modify it into a new set of solutions to fit your circumstances.
Three key discoveries of TRIZ were:
  1. Problems and solutions were repeated across industries and sciences
  2. Patterns of technical evolution were repeated across industries and sciences, and
  3. Breakthrough innovations used scientific effects outside the field where they were developed.

Why learn TRIZ?

TRIZ was formed to help solve technical problems without compromise. Solving a problem so that the problem disappears usually involves a highly inventive solution that is sometimes patentable. Here are some companies that are currently using TRIZ.

  • Samsung – When they first started using TRIZ, they had two consultants spend 8 weeks with a few of their engineers. They produced 50 feasible ideas and more than 10 patents. In 2003, Samsung used TRIZ on 67 R & D projects. They produced 57 patents and estimate saving $150 million in development time and energy. Samsung’s R & D budget was $2.9 billion and they had 4000 people trained in the methodology.
  • Proctor & Gamble – From 1985 to 1994, P & G produced less than 200 patents/year. Since introducing TRIZ in 1994, their patent output has grown by 100 patents/year. In 2000, they produced almost 600 patents.
  • HP has several hundred people trained in TRIZ.
  • Delphi has integrated TRIZ into all of their Six Sigma efforts.
  • 3M was a heavy user in the past and, after a period of stagnation involving a corporate reorganization, is again becoming a significant user.

People who would benefit from learning TRIZ are:

  • Researching Scientists – The regular TRIZ Inventive Problem Solving tools will help you to come up with innovative solutions faster. The goal is not to compromise. The goal is to make the problem disappear.
  • Design and Development Engineers – The regular TRIZ Inventive Problem Solving tools will help you break walls in the development effort. In addition, Technology Roadmapping will help you to see the next step in technological advancement and get a head start in its development and design.
  • Intellectual Property – A growing number of companies are starting to use TRIZ Technology Roadmapping to invent new technology and build patent fences around it. Money is then poured into R & D and, using Inventive Problem Solving, the company creates a product that sweeps the market. Even more troubling, companies are using TRIZ Inventive Problem Solving to circumvent and break patents. By using TRIZ to analyze patent applications, inventors and intellectual property lawyers can bulk up their claims, helping to protect the company’s IP. In some cases, the analysis comes up with additional inventions and more patents.
  • Managers – This method is a powerful tool for solving any type of problem. There are many examples where it has been adapted for use in business and social situations. TRIZ tools are useful as the problem resolution part of the Theory of Constraints and Lean Six Sigma. Lean and TOC identify and prioritize your problem. TRIZ assists in solving the stubborn ones.

Micro-Intro to TRIZ

This will give you a quick intro to the main problem solving algorithm.

What is the Theory of Inventive Problem Solving?

The Theory of Inventive Problem Solving (TRIZ) is a powerful algorithm for deriving highly innovative solutions to tough problems. TRIZ development was started in Russia by Genrich Altshuller. Altshuller analyzed 40,000 patents for the underlying principles behind the inventions. This study started the “science of invention” that was to be his life’s work.
The TRIZ algorithm has two main parts. The first part consists of thoroughly defining a problem. The second part uses TRIZ tools to develop innovative solutions.

The TRIZ Algorithm

  1. Define the Problem.
  2. Determine the Ideal Final Result.
  3. Create a Functional Model of the Problem.
  4. Identify Harmful Functions to Trim.
  5. Find Your Resources.
  6. Use The TRIZ Tools.
  7. Evaluate, Combine, and Implement Solutions.

45% of the time, your problem will be solved before you begin using the TRIZ tools in step 6. This is because the TRIZ algorithm demands a complete understanding of the problem before any thought of solutions takes place. This goes against the grain of typical problem solving. Most of the time, you see a problem and immediately start thinking about ways to fix it. Not much systematic thought is given to the reason for the problem. A TRIZ trained problem solver spends most of his or her time meticulously mapping the entire anatomy of the problem. Once mapped, the problem solver is then in the position to propose several possible solutions to the different parts of the problem.

Major TRIZ Tools

  1. Technical Contradiction Analysis – If you can formulate your problem into a contradiction where one improving feature is met with a worsening feature, you may be able to obtain new directions for innovation from the Contradiction Matrix.
  2. Physical Contradiction Analysis – For an inherent contradiction, one in which an object must perform two opposing functions, you can apply the “four ways” to remove them.
  3. Patterns of Evolution – This tool helps inventors see how their product will evolve. Such insight allows one to pursue large leaps in improvement. There are eight patterns of evolution that technological systems follow. The Patterns of Evolution are a key component in Technology Roadmapping.
  4. Scientific Effects – Once you have generalized your problem, you can then go out and look for solutions that do the same thing in other industries.
  5. Substance Field Analysis – Structures your system as two or more substances operated on by one or more fields. Once composed, Su-field Analysis provides a way for the user to replace substances and/or fields to remove problems. By doing Functional Analysis in Step 3, you will have performed a version of Su-field Analysis.

The Bloodbath to Come

All these lead paint toy scandals are only the beginning. On July 12, 2007, Businessweek ran a story, “Made in China: Faulty Tires.” This story foreshadowed the bloodbath to come. For every scandal, we can expect several high profile lawsuits to follow. On this blog, we’ve been talking about how American businesses encouraged the bad system that allowed for all these scandals to happen and keep happening. It is a good thing our legal system sucks, too. These businesses are now about to feel the pain of a predatory system that loves to take advantage of scandals for profit. In fact, it looks like it has already begun:

  1. Mattel hit with lead paint class action suit
  2. Couple files suit over lead in toys
  3. Toy Lawsuits Push Debate

I’m not even going to mention the many, many toy lawsuit websites that have been thrown up by ambulance chaser law firms trying to score a buck off these recalls. Personally, I’m really torn. I don’t know who to root for. The business that messed up and poisoned my kid with lead or the predatory lawyers that are going to steal millions from these companies. I guess there is no one to celebrate in this case. Waste begets waste begets waste. If only someone had the decency to do the right thing in the first place, we’d be spending our money on food instead of lead detector kits. I guess Enron and Worldcom really didn’t teach us anything. How about a new blog label called, “Stupid?”