Share This

Wednesday, February 10, 2010

The Fable Of Market Meritocracy

Markets don't reward smart people, nor should they.

It's a good thing that French President Nicolas Sarkozy is brimming with amour propre because he certainly did not earn any amour from the business elite gathered in Davos last month. In a bombastic riposte--delivered, no doubt, in one of his fabulously expensive designer suits-- he proclaimed that the recent financial meltdown had demonstrated that letting markets decide executive compensation was "morally indefensible." "There are remuneration packages that will no longer be tolerated because they bear no relationship to merit," he said.

But here's some news for Mr. Sarkozy: Markets don't reward merit; they reward value--two very different things. If Mr. Sarkozy does not appreciate the difference, it's not his fault actually. Most advocates of markets have failed to fully make this distinction, perpetuating a cult of market meritocracy--something that has hindered, not helped, the cause of free markets.

With the notable exception of Nobel laureate F.A. Hayek, market theoreticians have to a large extent employed the equivalent of the Great Man theory of history to explain what makes markets tick. According to this theory, the course of history is shaped not by the convergence of multiple, unpredictable events but by the intervention of great men. Likewise, in the conventional thinking about markets, economic progress depends not on the labors of infinite economic actors but on the select few, the brainiacs, who rise to the top and generate innovations from which ordinary mortals benefit through a kind of trickle-down effect.

English sociologist Michael Young noted in his influential 1958 fable, The Rise of the Meritocracy: "Civilization does not depend on the stolid mass, the homme moyen sensuel, but upon the creative minority, the innovator ... the brilliant few … the restless elite who have made mutation a social as well as a biological fact." Less elegantly, Ayn Rand evinced a "pyramid of ability" in capitalism under which "the man at the top contributes the most to all those below him." What's more, this Nietzsche of capitalism opined: "Man at the bottom who, left to himself, would starve in his hopeless ineptitude, contributes nothing to those above him, but receives the bonus of their brain."

What's good about markets in this line of thinking is that they identify the incandescent geniuses among us and catapult them to the top where their innate brilliance is harnessed to improve the lot of mankind. At once, then, markets yield economic progress and what Rand (and others) regard as justice--the biggest rewards to the best.

The only problem with this neat little formulation is that it is wrong at every level. For starters, the idea that value creation is a one-way street from the top to the bottom is not just offensive, but it ignores the principle of comparative advantage, a key breakthrough in market theory. Put simply, this principle holds that everyone benefits by exchanging goods and services with everyone else, regardless of anyone's inherent capabilities. It's in the interest of even the most annoying "all-rounder" (as we say in India), who is better than me at everything, to specialize in those tasks in which our gap is the biggest and trade with me for those in which our gap is smaller. Under the elaborate division of labor that ensues, both the less-endowed and the better-endowed contribute to each others well being.

But is it the case that this division of labor necessarily directs the biggest rewards to the most gifted by putting them at the highest end of the value chain? No.

The beauty of the market, Hayek brilliantly pointed out, is that they allow people to use knowledge of their particular circumstances to generate something valuable for others. And circumstances, he emphasized, are a matter of chance--not of gift. Furthermore, since no two people's circumstances are ever identical, every producer potentially has something--some information, some skill or some resource--that no one else does, giving him a unique market edge. "[T]he shipper who earns his living from using otherwise empty or half-filled journeys of tramp-steamers, or the estate agent whose whole knowledge is almost exclusively one of temporary opportunities, or the arbitrageur who gains from local differences of commodity prices, are all performing eminently useful functions based on special knowledge of circumstances of the fleeting moment not known to others," noted Hayek.

In a functioning market, Hayek insisted, financial compensation depends not on someone's innate gifts or moral character. Nor even on the originality or technological brilliance of their products. Nor, for that matter, on the effort that goes into producing them. The sole and only issue is a product's value to others. Compare an innovation as incredibly mundane as a new plastic lid for paint cans with a whiz-bang, new computer chip. The painter could become just as rich as the computer whiz so long as the savings from spills that the lid offers are as great as the productivity gains from the chip. It matters not a whit that the lid maker is a drunk, wife-beating, out-of-work painter who stumbled upon this idea through pure serendipity when he tripped over a can of paint. Or that the computer whiz is a morally stellar Ph.D. who spent years perfecting his chip.

The idea that there is no god (or some secular version of him) meting out cosmic justice through the market's invisible hand is unsettling, even to market advocates, but it shouldn't be. It opens up the possibility of a defense of markets that is, as it were, more marketable.

Few would dispute that markets are fairer than the aristocratic order they replaced where privilege was a birthright, not something to be earned. But the view that the super-gifted or the super-smart deserve the biggest rewards doesn't seem a whole lot fairer given that these traits are arguably inherited, too. This conception, in fact, forces those who are less successful to internalize their failure--accept their second-class status as preordained--breeding alienation and resentment. Hard work or some quality of character would offer a more palatable basis for building a case for markets, except that all the lowlifes who routinely make it rich in markets offer too much evidence to the contrary.

Hayek's understanding of markets overcomes these problems by, first and foremost, democraticizing the concept of merit. If anything in your possession, no matter how trivial--some local knowledge, some quirky interest--can potentially be turned into something useful for others, then there is not any one formula for market success; there are a potentially infinite number. This means that success is possible for a far wider range of people in a market, making market societies inherently less hierarchical than more closed ones.

Take, for instance, India in its preliberalization days. Economic opportunities were exceedingly limited in its regulated and centrally planned economy. The most sought-after professions were engineering, medicine, accounting and--hang on to your fountain pens!--civil service, because they offered a path to secure jobs in government-approved sectors. Competition for professional colleges was fierce. The lucky few who made it into elite institutions such as the Indian Institute of Technology for engineering were regarded almost as a special breed. Even now, the unabashed elite-worshipping that IIT graduates command in India would make Zeus blush.

But free markets change all this. They close the talent-gap by allowing people to ferret out and market whatever they've got--even, regrettably, Paris Hilton. In America, for instance, there are opportunities galore for funny people--standup comedy, late-night talk shows, etc.--who may have no head for math or science. Their sense of humor is a prized commodity, a gateway to riches and fame, instead of social ridicule as it would have been in the India of yore.

But markets don't just expand and democratize the concept of merit; they render it moot. No longer does it matter what great qualities reside in you. What matters is if you can make them work for others. The concept of merit is replaced by that of value. Merit is intrinsic, concentrated and atomistic; value is relational, decentralized and social.

The need for embedding this Hayekian understanding of markets in the public consciousness cannot be overstated. And the first step in doing so might be purging the word "merit" from the vocabulary of markets and replacing it with value. This would make it much easier to explain how no functioning industry, absent access to taxpayers' pockets, can afford forever to pay its executives obscene salaries beyond the value they are generating. At once, then, it would be possible to oppose both the recent government bailouts and government regulations such as Sarkozy-style caps on executive salaries.

More importantly, it would become possible to counter the popular perception--the source of so much hostility against markets--that there is some body of super-elites, masters of the universe, who can sit in their plush offices on Wall Street and Silicon Valley and reign supreme through their sheer brain power. If value, not brain power, is the engine that drives markets, then the market's inherent nature militates against elite control.

Markets are a fundamentally antielitist social force. If this is not generally recognized, it is not so much because of what the enemies of markets say to attack them--but what their friends have said to defend them. To rescue markets, then, one has to rescue them from their friends first. Mr. Sarkozy is not the main problem here.

Shikha Dalmia is a senior analyst at Reason Foundation and a writes a biweekly column for Forbes.

How to improve your investment skills

Personal Investing - By Ooi Kok Hwa

WE have been asked by many readers on ways to improve their investment skills. In fact, for all of us who invest, it is one of the essential skills that we need to acquire in our lifetime. Like it or not, we need to have it if we need to generate returns for our investment.

All investors want good returns from their investments. However, most of the times, instead of generating returns, retail investors are suffering from losses from their investments. We feel that one of the key differences between an intelligent investor versus a normal investor is that the intelligent investor will be aware that he may make mistakes in some of his investment decisions while a normal investor tend to overlook the fact that he will make wrong decisions no matter how good he thinks he is.

Despite extensive research on certain listed companies, due to some unforeseen changes in certain fundamental factors, even good value companies may suffer losses. Under such circumstances, an intelligent investor will admit that he had made a mistake in his investment decision and will cut losses fast.

However, the problem with most investors is that they refuse to face their mistakes; some are not willing to cut their losses even though they are aware of their mistakes.

Hence, rule number one in investing is that we must be fully aware that regardless of whether you are an investment guru or an average investor, everyone will make mistake in his investment decisions. That’s why some experts say: “When somebody mentions that they have more experience than you, they mean that they have incurred more losses than you in stock market.” The key is to learn from our mistakes.

In order to avoid incurring losses in stock market, we need to develop our own investing system that suit our needs, skills, knowledge and risk tolerance level. The investing system can be adopted from the fundamental analysis, technical analysis or combination of both. If we ask some remisiers, they will most likely tell you that they need two to three years to develop their own investing system that can help them to generate returns from stock market.

One of the fastest ways to acquire investing knowledge is through reading books relating to investment. There are many good investment books in the market. However, since every investor has different preferences, the best way is to visit bookstores and look for investment books that he or she can understand and can offer the skills needed. For beginners, always start with some basic investment books that explain well on key investment concepts.

Here are some good investment book titles for consideration: The Intelligent Investors (by Benjamin Graham), The Essays of Warren Buffett: Lessons for Corporate America (Warren Buffett and Lawrence A. Cunningham) and Rule #1 (Phil Town). For advanced investors, you may consider Security Analysis (by Benjamin Graham and David Dodd), which is still one of the best investment books in the world.

Apart from reading books, investors need to read more business news in newspapers and magazines to keep themselves updated on the latest happenings. In addition, many newspapers, magazines and websites also publish good articles for the purpose of educating general public on investment. For example, investors can get good investment knowledge from website like www.min.com.my, by Securities Industry Development Corp.

Reading analysts’ research reports will enhance our understanding on some issues and factors in valuation as well as comments on some corporate strategies and developments. This knowledge is crucial in helping us making better investment decisions. Besides, for those serious fundamental investors, they may consider buying books like Stock Performance Guide (by Dynaquest Sdn Bhd) and Shares (Pioneers & Leaders (Publishers) Pte Ltd), which will provide all the essential investment information like companies’ background and some key critical investment information.

Another way to acquire investing knowledge is through attending investment training classes. There are many types of investment training classes, for example, classes on fundamental investment, technical analysis, currency trading or option trading. Given that a lot of these classes are quite expensive, we need to check whether investment training suits our needs. We believe some of those classes may be able to help investors generating returns, however, they require higher level of discipline and commitment.

Before we start investing with “real” money, one of the ways to gain experience and at the same time test out our skills is by building up a “virtual” portfolio and investing using “virtual” money. We can always try out our investment skills through playing a simulated investment game and monitor the investment returns before putting the real money into the stock market. Besides, we should also start young. If we acquire these investment skills at younger age, the losses that we may incur will be much lower than trying them when we are getting nearer to our retirement age.

● Ooi Kok Hwa is an investment adviser and managing partner of MRR Consulting

How to Improve Your Property Investment Buying Skills

Investing in real estate is your first step towards a productive life. So if you want to rack up large sums of money without having to face big risks, the first thing you should do is to go on a property investment
buying spree.

Despite the current economic crisis, real estate investing has never lost its appeal among many people. In fact, the number of real estate investors continues to grow and it looks like this trend is likely to last for the next few years. However, because more and more people are seeking to invest in real estate, you should learn how to outsmart the competition for you to succeed in this business.

One way to boost your knowledge of real estate investing is to do some research. As we all know, education is important in our lives because it teaches people new things and boosts their skills. So if you want to become a great real estate investor, you shouldn’t stop educating yourself just because you have managed to land a great deal on your first try.

A good strategy to improve your knowledge of property investment buying is to join the local real estate investing association or club. Setting the competition aside, befriending your fellow real estate investor can do a lot for your career. By surrounding yourself with individuals who share the same interests as yours, you will be able to obtain valuable information that you wouldn’t learn in books.

If you are truly interested in investing in real estate, then you should know the latest happenings in the housing business. Browse newspaper or surf the Internet to keep yourself informed of the latest market trends. This will help you make better decisions when buying investment properties.

The World Wide Web is also a treasure trove of ideas and information about real estate investing. There are numerous websites that offer valuable tips about buying investment properties. Many real estate gurus are also sharing their knowledge online with those who are just starting out in the business. So if you want to become better at what you do, just launch a browser and explore the Internet.

Meanwhile, there is a website the offers quality real estate education for those who want to learn more about property investment buying. REIWired.com is home to dozens of nifty articles, sound files, and videos that can bring out the great investor in you. So if you want to feed your brain with quality real estate content, visit REIWired.com.
Retrieved from "http://www.articlesbase.com/business-articles/how-to-improve-your-property-investment-buying-skills-1412870.html"

Tuesday, February 9, 2010

Different Colors Describe Happiness vs. Depression

By Stephanie Pappas, LiveScience Contributor

Color Wheel for mood:http://i.livescience.com/images/mood-color-wheel-100208-02.jpg
People tend to pick different colors to describe different moods, according to research published Feb. 8, 2010 in the journal BMC Medical Research Methodology.

Are you in a gray mood today? How about a blue funk? Maybe you're seeing red, because you're green with jealousy. The colors we use to describe emotions may be more useful than you think, according to new research.

The study found that people with depression or anxiety were more likely to associate their mood with the color gray, while happier people preferred yellow. The results, which are detailed today in the journal BMC Medical Research Methodology, could help doctors gauge the moods of children and other patients who have trouble communicating verbally.

"This is a way of measuring anxiety and depression which gets away from the use of language," study co-author and gastroenterologist Peter Whorwell of University Hospital South Manchester told LiveScience. "What is very interesting is that this might actually be a better way of capturing the patient's mood than questions."

Colors are often used as metaphors for moods, but no one had systematically researched color associations, Whorwell said. To investigate, he and his colleagues picked eight colors — red, orange, green, purple, blue, yellow, pink and brown — and split each into four shades. They then added white, black and four shades of gray for a total of 38 options. After meeting with focus groups, the researchers decided to display the colors in the form of a wheel.

Next, they recruited 105 healthy adults, 110 anxious adults and 108 depressed adults and mailed them printouts of the color wheel. Each person was asked to pick their favorite color, as well as the color they were most "drawn to." Finally, they were asked to pick a color that described their day-to-day mood over the last several months. Another group of 204 healthy volunteers classified each color as positive, negative or neutral.

Whether depressed, anxious or healthy, people liked blue and yellow. Blue 28 on the color wheel was the most popular favorite color among healthy people, while Blue 27 (which is a little darker than 28) got first place among people with anxiety and depression. Meanwhile, Yellow 14 was picked as the color most likely to catch the eye.

But when it came to mood, the groups diverged. Only 39 percent of healthy people associated their mood with a color at all. Of those who did, Yellow 14 was the most popular choice, with about 20 percent of the votes. Meanwhile, about 30 percent of people with anxiety picked a shade of gray, as did more than half of depressed volunteers. In comparison, healthy volunteers described their mood with a shade of gray only about 10 percent of the time.

The researchers also found that when assigning a mood to colors, saturation matters.

"A light blue is not associated with a poor mood, but a dark blue is," Whorwell said. "The shade of color is more important than the color itself."

Whorwell is now testing the wheel on patients with irritable bowel syndrome. He's hoping that color choices can reveal patients' attitudes and predict how well they will respond to treatments like hypnosis. Because people are embarrassed by gastroenterogical symptoms, Whorewell said, non-verbal methods of getting information are sometimes preferable to conversation. And, he said, with additional research, the wheel could be used in medical fields from pediatrics to surgery.

"You've got an instrument now," Whorwell said. "Now people have to play with it and find out the applications."

* What Color Fits Your Mood? http://www.livescience.com/php/multimedia/imagedisplay/img_display.php?s=health&c=news&l=on&pic=mood-color-wheel-100208-02.jpg&cap=People+tend+to+pick+different+colors+to%0D+describe+different+moods%2C+according+to+research+published+Feb.+8%2C+2010+in%0D+the+journal+BMC+Medical+Research+Methodology.&title=

* 7 Thoughts That Are Bad for You: http://www.livescience.com/health/090911-7-bad-thoughts.html

* Top 5 Keys to Happiness: http://www.livescience.com/health/080822-top5-keys-happiness.html

The High Cost Of The U.S. Budget

David Malpass and Eric Singer, 02.08.10, 03:10 PM EST

Remember the ending of ''Thelma and Louise''? This is worse.

Early last week the administration unveiled its new $3.8 trillion budget. This budget, and the spending it assumes, is so reckless it is difficult to fathom. It projects a deficit of $1.6 trillion for fiscal 2010, with an explosion of publicly traded government debt to $18.5 trillion by 2020. Debt service is projected to reach $912 billion a year by 2020, or over $5,000 for every working person in America.

Even assuming that wages grow over time, the budget suggests the government believes that as U.S. citizens we should devote 10% or more of our grossincome just to pay the interest on the federal debt. That 10% comes out of income even before people start paying for actual government programs like national defense, health care, bank rescues and welfare. Nor does it count the spending people need to make for state and local government obligations and their own mortgage and credit card payments.

The administration clearly believes that we can spend our way out of a slow economy and has put us on a path to have total government spending over 40% of GDP. This is Disneyland thinking. While it is true that we can spend our way into a depression, there is little evidence we can spend our way out. It didn't work last year, when we allocated $ 787 billion in stimulus spending on the threat that unemployment would go over 8% otherwise. (It is now at 9.7%--16.5% if you count the underemployed and those who are too discouraged to look for work.) Excess spending didn't work for President George W. Bush. It didn't work for Presidents Roosevelt or Obama.

When the government spends $3.8 trillion, there is no net multiplier effect. There is a divider effect. We become divided from our income and wealth. The government taxes inefficiently and spends inefficiently, wasting a big chunk of the transfer. Yes, it is possible for purposes of a speech anecdote to find isolated beneficiaries. The government will soon be able to claim massive "job gains" for temporary workers for the 2010 census. But much government spending misallocates resources--it's distributed without a profit motive and with frequent conflicts of interest, including an eye toward political returns rather than longer-term growth and private sector jobs.

Markets are waking up to the fiscal crisis. The price of a credit default swap for a five-year U.S. Treasury has gone up 700% since 2007, from about 8 basis points to 56 basis points. Germany, at 40 basis points, is now deemed by the market to be 28% less likely to default than the U.S., and the difference is growing. More ominously, the U.S. swap has more than doubled in price in the last few months. As recently as 16 months ago Greece was at levels we are today. When a country goes past owing 100% of its nominal GDP in publicly held debt, as this budget schedules us to do by 2020, there is danger.

Given our politicians' willful innumeracy, perhaps the best way to understand what the budget does for us is to compare it with the movie Thelma and Louise. For those who never saw it, Thelma and Louise was a film about two women on a supposedly liberating road trip. Toward the end, running out of money, they rob a convenience store and blow up a gasoline truck for fun. FBI agents close in on them, and they decide to drive their car off a cliff. In a Hollywood ending they are shown with a look of joyous freedom while they are in free fall.

Like Thelma and Louise, we are spending beyond our means, and at some point the bond vigilantes will close in. With this budget we almost guarantee that the United States AAA debt rating will at some point be called into question. If and when it is, the result will resemble the downward arc of that terrific 1966 Pontiac Thunderbird. In real life, at the end, when you go off a cliff at 100 miles an hour, gravity takes hold, and there is a fiery crash.

David Malpass is an economist and president of Encima Global LLC. He writes a column for Forbes. Eric Singer is the president and portfolio manager of the Congressional Effect Fund (CEFFX).

Monday, February 8, 2010

IBM Unveils Power7 Systems, Ramps Up Competition Against HP, Sun

IBM is rolling out the first of its Power7 systems as it looks to grab more market share in the $14 billion Unix server space from HP and Sun. The move comes as HP and Intel prepare to release the next-generation Itanium processor and Oracle looks to bring Sun into the fold.

IBM is rolling out the first systems based on its new Power7 processors, setting the stage for a new round of competition in the rapidly changing high-end server space.

IBM will officially unveil the four new Power7 systems at an event Feb. 8 in New York. At about the same time, at a press conference in San Francisco that also will be Webcast, officials with Intel and Hewlett-Packard are expected to officially release “Tukwila,” the much-delayed next-generation version of the Itanium processor.

All that comes as Oracle works to incorporate Sun Microsystems’ SPARC/Solaris hardware line into its business, and Intel and Advanced Micro Devices look to grab more high-end workloads with their x86 processors.

“We’re seeing three very big vendors … sharpening their elbows a bit,” Charles King, an analyst with Pund-IT Research, said in an interview.

IBM officials are looking to press their advantage, having gained 12 points of market share in the $14 billion Unix space since 2005 and more than 2,200 HP and Sun server and storage customers over the same time span.

“We view the Unix market as very, very robust and mission-critical, and with [the Power7] platform, we have a perfect spot in the market,” Scott Handy, vice president of worldwide marketing and strategy for IBM’s Power systems platform, said in an interview.

With the new Power7 systems, IBM is offering greater performance while consuming less energy than their predecessors, according to Handy. The servers also dovetail with IBM’s Smarter Planet initiative, with capabilities to not only process large amounts of data, but also to analyze that data at the same time.

According to IBM, the Power7 servers deliver twice the performance and four times the virtualization capabilities for the same price as the Power6 servers, all the while consuming half the energy.

They also offer better price for performance than comparable systems from Sun and HP.

The new Power7 chips bring greater performance than the Power6 processors. Power7 offers up to eight cores, with each core able to run up to four instruction threads, enabling each chip to run 32 simultaneous tasks. That includes four times the number of cores and eight times the number of threads per chip than in Power6.

However, the new servers come with more than just new chips, Handy said. There are a host of new integrated hardware and software features designed to increase performance and energy efficiency, Handy said.

TurboCore mode is a workload optimization feature that can have four cores running, and putting the resources—include cache memory and memory bandwidth—of the other four dormant cores before those active ones. It also can increase the clock speed of those four active cores.

Not all systems will support TurboCore mode. Among the four new servers, only the Power 780, designed for such high-end transaction workloads as databases, will support TurboCore.

When not in TurboCore, all the Power7 systems are in MaxCore mode, which takes advantage of its increase in thread count.

The chips also feature Intelligent Threads, a technology that can dynamically change the number of threads being run depending on the type of workloads.

Another Power7 feature, dubbed Active Memory Expansion, uses memory compression technology to make the physical memory appear to be twice as large as it actually is, and can dynamically change the amount of compressed memory based on workload demands.

In addition, IBM engineers have optimized the company’s middleware—including WebSphere, DB2 pureScale, Lotus Domino and Rational—to take advantage of the new capabilities in the Power7 systems.

Regarding virtualization, Power7 systems can support 1,000 virtual machines on a single physical server, four times as many as Power6 servers can support.

“They can put lots of virtual images on these machines and really consolidate the number of servers,” Handy said.

In addition, IBM’s Intelligent Energy technology lets customers power on and off parts of the system, or ramp up or down processor speed based on thermal conditions and system utilization, he said. Intelligent Energy can be run on a single server or across multiple systems. The goal is to help users balance the competing needs of energy consumption, performance and utilization.

The gains in performance and energy efficiency enable IT administrators to do the same amount of work on fewer systems, or to do much more work on the same number of servers, Handy said.

Along with the Power 780, IBM is rolling out the Power 770, a midrange server with up to 64 cores; the Power 755, a high-performance cluster with 32 cores; and the Power 750 Express, a midmarket offering.

IBM also is enhancing its Systems Director Express, Standard and Enterprise editions to make management easier and to bolster virtualization capabilities through enhancements to VMControl.

Power7 - Big Blue eye on UNIX Four server preemptive launch

By Timothy Prickett Morgan • Get more from this author

The scuttlebutt is that IBM seemed perfectly content to wait until May to launch the Power7-based Power Systems servers, but something changed and compelled the company to move up the announcement of its first machines using the eight-core processor to today. Big Blue is not in a habit of explaining its motives or its timing for product launches, but it seems clear that IBM wanted to get out in front of a whole lot of processor and systems launches that are expected between now and the summer.

With so many customers expecting Power7-based machines, it wasn't like IBM was going to have stellar sales of existing Power Systems machines, which run the AIX, Linux, or i/OS operating systems and which are based on the Power6 and Power6+ processors.

The machines announced today are clearly aimed at blunting the attack of midrange X64, Itanium, and Sparc servers as well as some bigger boxes that are going to start creeping up into the power class of the current top-end Power6-based Power 595 machine, which packs 64 cores running at 5 GHz into a single system image. IBM is especially focused on the Unix part of the Power Systems business, where it has generated $1.6bn in revenues in takeouts of Hewlett-Packard and Sun Microsystems boxes in the past three years.

Scott Handy, vice president of worldwide strategy and marketing for the Power Systems division at IBM, says that Sun and HP are each bringing in about $4bn a year in Unix systems sales, with Big Blue getting the lion's share of what is left over of the $14bn Unix market.

"We have 40 per cent share, but this is still a tremendous opportunity here," says Handy. "We are taking Unix to a new leve." And by Unix, IBM apparently means both AIX and i/OS, which share the scalability, reliability, and energy-efficiency attributes of a system designed to support mission-critical workloads.

"We want to position Power as the future of Unix. HP and Sun haven't caught up to Power6, and we will trounce them with Power7. The best marketing executive in the world cannot position Tukwila against Nehalem-EX, which does not support HP-UX," Handy says with a laugh, talking about the four-core Itanium chip that Intel will announce today and the eight-core Xeon chip due sometime in the first half of this year. "HP-UX customers are going to be just as distraught in 2010 as Sun customers were in 2009."

Neither HP nor Oracle, the new owner of the Sun Solaris server business, seem to think there is as much blood in the water as Big Blue thinks it smells. But clearly there is going to be some fierce competition in the server racket this year, particularly with so many projects delayed since the summer of 2008, leaving servers in need of capacity upgrades, and budgets under continuing pressure.

The blood may have more to do with red ink associated with steep discounting than technical superiority when all is said and done in 2010. Which would make it look a lot like 2002 and 2003 - truly awful years for Sun, so-so for the then-merged HP/Compaq, and great ones for IBM's Unix biz.

Contrary to the rumors from last week, IBM did not launch the Power7 line with a kicker to the current Power 520 entry server. (More than a few readers told me this in private; perhaps they all heard the same wrong information). What IBM is rolling out today are four machines in the midrange and enterprise class, which leave entry and blade servers as well as big iron boxes to come out later in 2010.

Handy would not be specific about exactly what these machines would look like, but after some arm twisting he said that customers with current Power 520s (presumably using Power6 and Power6+ chips) will have an upgrade path to the future Power7-based entry machines (presumably to be called the Power 720 to be consistent with the names chosen for the four machines launched today).

Handy also confirmed that the kicker to the high-end Power 595 would have the same 32 sockets, presumably to be called the Power 795, and added that it would cram 256 cores into the same thermal envelope that a Power6-based machine with 64 cores had.

The mount

Like the prior generations of Power-based machines from IBM, the CPUs are mounted on processor cards that mount vertically into the system motherboards. Each processor card has sockets for the Power processors and slots for the main memory. I/O subsystems and external peripherals plug into the system board, which links out to the processor card through a backplane. (Generally speaking, this is the same sort of design that is expected with Intel's future "Becton" Nehalem-EX Xeon processors for multiprocessor SMP machines).

Like IBM's high-end System x servers, which can scale outside of the box using NUMA-like clustering (thanks to some technology Big Blue got when it bought Sequent Computer Systems in July 1999 for $180m). Up to four Power server nodes can be linked into a single system image for AIX, i/OS, or Linux using this interconnect. This interconnect is not available on every machine, and customers pay a premium for the scalability that comes from this interconnect.

The first new machine in the Power7 family of systems is the Power 750, which is the follow-on to the midrange Power 550 that was based on the Power5 and Power6 generations of chips. The Power 750 comes in the workhorse 4U chassis that IBM has had since 2004, but it has been modified to allow for up to eight front-mounted, small form factor disk or SSD drives. That 4U chassis can be configured as a rack or tower server and has room for up to four processor cards, each with 16 DDR3 main memory slots and each with a single processor socket.

This box has the widest variety of Power7 processor options among all of the machines announced thus far. One option has six-core Power7s running at 3.3 GHz, another uses eight-core processor cards that run at 3 GHz or 3.3 GHz. There is a top-end machine that only comes with four processor cards with all 32 cores turned on and running at 3.55 GHz. There seems little doubt that this box will carry the heftiest price.

The Power 750 offers from 8 GB to 512 GB of main memory expansion, with a maximum of 128 GB per processor card using 8 GB DDR3 DIMMs. The box has three PCI-Express and two PCI-X slots peripheral slots and a single GX+ adapter slot, which is used to hook remote I/O drawers (based on a modified InfiniBand link that IBM calls 12X) or graphics cards into the machine. The Power 750 can have four 12X I/O drawers using PCI-Express slots or twice as many using PCI-X drawers. With all of the I/O drawers in the box, the Power 750 can have 584 disk or SSD drives directly attached to it.

The Power 755 is a version of the Power 750 server that is tweaked specifically for supercomputing workloads (and maybe for parallel database clusters running IBM's DB2 PureScale software, but the company has not said). The Power 755 only comes with the four processor cards in the box loaded with 3.3 GHz cores and all 32 cores are activated. This machine can only have 64 GB of memory per processor card, however, which means memory tops out at 256 GB. Other than that and the fact that this machine only supports AIX and Linux, it looks just like a Power 750.

The 770 and the 780

The Power 770 is a version of the Power 750 chassis that has two disk bays removed and the NUMA/SMP clustering bus added in. The Power 770 server node also has two GX++ slots per node, which means it can support a lot more 12X I/O drawers and therefore a lot more peripherals. Each Power 770 server node can have six disks or SSDs and has six PCI-Express slots. With the maximum of 16 12X-based I/O drawers, the machine can support 184 PCI-Express slots across four server nodes and 1,320 disk or SSD drives.

The Power 770 nodes have two flavors of processor cards: one using six-core Power7s running at 3.5 GHz (with 12, 24, 36, or 48 cores activated) and another using eight-core Power7s running at 3.1 GHz (with 16, 32, 48, and 64 cores activated). Main memory on the machine spans from 32 GB to 2 TB. To get memory capacity up to 1 TB means dropping the memory speed from 1.33 GHz down to 1 GHz, and going up to 2 TB means dropping down to 800 MHz. For some workloads, the lower speed of the memory will negate the extra capacity benefits.

The Power 780 is essentially the same machine as the Power 770 with three changes. First, it uses processor cards that have two sockets per card instead of one, which means it has double the processor cores. (However, the memory slots stay the same at 16 per processor card and the maximum main stays at the same 32 GB to 2 TB range per machine).

The second change is that it comes in the same enterprise-class, green-striped chassis that the Power 595 and System z mainframes have. (It has the skin of a mainframe, but the guts of a NUMA cluster with an architecture that IBM has been developing for a decade and selling for six years.) The other big change is the machine only has one set of processor cards, but they have two operational modes. In MaxCore mode, as IBM calls it, each node in the Power 780 cluster has four processor cards for between 16 and 64 cores running at 3.8 GHz, depending on if you buy one to four nodes.

If you happen to have a workload, like an OLTP system, that would do better having fewer cores running at a higher speed, you flip a switch in the microcode, reboot the system, and when it starts up you run in TurboCore mode. In this mode, the cores run at 4.1 GHz, but only half of them turn on. Those remaining cores in the system have access to both memory controllers on the Power7 chips and its full 32 MB of embedded DRAM L3 cache memory. For database workloads, TurboCore mode can boost performance by 20 per cent over MaxCore mode on the same physical machine.

IBM is not publishing pricing information on the machines yet, but Handy says that the memory price cuts that IBM made on Power6 and Power6+ systems back in November, where it cut DDR2 memory tags by between 28 and 70 per cent, were in fact setting the prices on older memory to the same level as IBM expected to charge for DDR3 memory on the Power7-based servers.

As for the systems themselves, IBM's plan is to hold prices roughly the same as the prior Power6 and Power6+ systems and give customers the extra performance. "This is very aggressive price/performance for us, and we are striking while the iron is hot," Handy says.

All of the Power7 machines announced today can support the current AIX 6.1 release as well as the earlier AIX 5.3 release. Customers using i/OS are going to have to move up to the i 6.1.1 interim release that was announced last quarter, and to fully exploit the Power7 feature set, they will have to wait to see i 7.1 later this year. (The word on the street is that i 7.1 will be available in the third week of April, but IBM did not confirm this).

Red Hat Enterprise Linux is not supported on these Power7 machines, which is a bit odd, but apparently IBM and Red Hat are working on it. Novell's SUSE Linux Enterprise Server 10 SP3 and 11 will run on the machines today. IBM plans to start shipping the Power 750 and 755 servers on February 19, The larger Power 770 and 780 machines will ship on March 16. ®

IBM Powers Up For Server War

Andy Greenberg, 02.08.10, 12:00 AM EST

As the Oracle-Sun combo heats up hardware competition, IBM's Power7 servers aim to dominate.

Oracle and IBM's knife-fight over high-end enterprise hardware is about to begin--and IBM intends to bring a cannon.

IBM ( IBM - news - people ) plans to announce Monday its first Power7 servers, a long-awaited line of Unix machines that will integrate new chips and software. Big Blue is initially only releasing its mid-range Power7 servers, with both higher- and lower-performance machines to follow later in the year. But IBM says that even the mid-range Power7, the result of three years and $3.2 billion of research and development investment, will squeeze 64 processing cores into each server. It will also double the speed of each of those cores while quadrupling their power efficiency.

IBM's Ross Mauri, manager of the company's Power group, says the servers will be aimed at its "smart infrastructure" customers--those doing real-time analysis of vast amounts of data to optimize complex systems like traffic or water and electricity distribution.

"We're making hardware that was designed to meet the massive scale of these 'smarter planet' projects," says Mauri. One test customer over the last year has been San Mateo, Calif.-based eMeter, which will resell Power servers to manage the data flowing from its Internet-enabled electricity meters in consumers' homes.

But as Oracle ( ORCL - news - people ) completes its integration of Sun Microsystems ( JAVA - news - people ) and aims Sun's high-end servers at IBM customers, Forrester Research analyst Brad Day says a more important factor may be Power7's ability to compete with a wide range of other servers, both low-end x86 machines and high-end Unix hardware.

Unlike Power6 servers, Power7 machines can do the sort of fast-paced transaction processing needed by financial firms, complex analytics from simultaneous data sources or perform high "throughput" computing, including crunching the data-intensive models used in simulations like genomics or weather prediction. "This is the first time they've designed a system that's equally competitive in all these different workloads," says Day. "I'd say they're 18 to 24 months ahead of their competition."

IBM's Mauri pits those versatile machines head-to-head against Sun's offerings: "Sun has their Niagara line, which is throughput computing with lots of cores. But if you want to run transactional computing, they'll try to sell you SPARC," he says, referring to another line of Sun servers. "We have one architecture that's heads-and-tails above both those systems."

Big Blue's new systems aren't cheap: The servers IBM is initially releasing start at $30,000 in their simplest configuration and range up to $1 million.

But those machines likely will help IBM solidify its hold on the high-end server market. The company holds about 40% of market revenues for Unix servers, compared to less than 30% for its biggest competitors, Sun and Hewlett-Packard ( HPQ - news - people ), according to tech tracker IDC. That's a massive switch since the beginning of the last decade, when IBM had less than 20% share and Sun held close to 36%.

Part of that success has been IBM's willingness to foot the bill for companies to migrate their software from other platforms to its Power systems, and even offer discounts based on the value of their competitors' legacy hardware. "They're essentially putting a bounty on Sun and HP," says Forrester's Day.

Given its success at the high end of the hardware spectrum, IBM's larger threat may be the inevitable improvement of commodity servers built on low-end Intel ( INTC - news - people ) and AMD x86 chips.

But Day says that IBM's virtualization software, what it calls PowerVM, allows customers to consolidate their physical x86 servers into virtual machines that can be packed into one much larger, more efficient machine.

By virtue of that software and Power7 servers' sheer size, the new systems could give IBM a leg up on both its low-end and high-end competition. "Although it competes mainly with HP and Sun's high-end servers, Power7 allows them to fire missiles into the x86 space," says Day. "You can clear out a hundred x86 servers and replace them with three of these."