Archive for March, 2009

Dell Gets IT

The world is changing. It has changed. No longer can you keep up with the Jonses doing what you have been doing. You have to change too.

Dell will be pushing PCs to farmers in China, a huge emerging market. Dell will have to earn its keep selling low-cost PCs to people not locked-in to that other OS. I guess that means GNU/Linux will become more popular in a land where it is already popular. The government of China is subsidizing these purchases to boost the economy internally. The cities have been enjoying the fruits of the technology/manufacturing boom in China. Now the rural parts will join in.

This is a market of hundreds of millions of people now able to buy their first PC. The world will be tripping over themselves to sell and price/performance will matter and monopoly will have no place. Dell will likely be selling PCs made in China from Chinese parts with Chinese labour. Will Dell recommend Vista? I doubt it.

- Robert Pogson

The Sky is NOT Falling in IT

There have been many articles on the web lately describing a reduction in spending in IT this year. For example, IDC found 4.9% reduction in spending on servers in APEJ 2008/2007. This is tiny compared to the usual growth and the downturn in other industries.

Many are in panic, laying off workers and blaming the economic downturn for their troubles but there other many other forces at work. There must be some positives limiting the downturn in IT. If we consider Moore’s Law brought us cheap storage/memory/quad-core CPUs recently, and people have been seriously consolidating servers and services, the growth in spending is absolutely wonderful. The purpose of IT is to do everything better/faster/cheaper. We should not be upset that it is working and that the world can get done what it is doing with fewer servers/staff.

Another force to be reckoned with is the move to FLOSS. Servers running GNU/Linux can do more. Once we have enough servers idling for redundancy and load balancing, perhaps we do not need so many. The slowdown will be affecting startups greatly and established outfits less. Instead of panic, we should see these changes as a normal part of IT and adjust.

Unfortunately for GNU/Linux on the desktop, many OEMs seem to have reduced production of systems with GNU/Linux at the very time when they could move a lot of product. By sticking with M$, they will harm themselves as well as their customers by prolonging the monopoly. The OEMs who push GNU/Linux sincerely will make a killing in emerging markets at the low end of unit prices. Those sticking with M$ will not be able to compete on price no matter the subsidy they get form M$. The layoffs at M$ show us that the cash cow is drying up.

We have similar declines in the desktop PC as notebooks become more popular. Is the solution to stick with M$ and decline with them? No! OEMs should produce less expensive desktops based on mini/micro/nano ATX and low power CPUs, probably thin clients. If they make them as cheap as the Chinese are, they will sell and they will have growth in desktops, not decline because a tiny desktop is better than a notebook except for portability. Few of us are traveling salesmen. Many of us can log in at a different terminal in the organization instead of carrying a burden with us.

Moore’s Law is expected to permit more computing in a small package for less money. OEMs who try to keep prices up instead of increasing volume are missing the boat. They should not prolong or intensify the downturn by fighting Nature. Let GNU/Linux on low-powered devices and others thrive. The world is demanding lower-powered CPUs for clients. Intel, AMD, VIA, etc. should give the world what the world wants and there would be more growth. Lower unit prices are inevitable. Decline in the industry is not.

- Robert Pogson

The Avalanche of the Crumbling Monopoly

Psystar is counter-suing APPLE for the freedom to install MacOSX on non-Apple hardware. Whoops! If they win, everyone will be able to install MacOSX on PCs. That will take another bite out of the monopoly as huge numbers of PCs could go that way. Now MacOSX is restricted to APPLE’s machines. Opening that up will give consumers a straight-up choice of OS:

  • that other OS
  • MacOSX
  • GNU/Linux

Whatever increase in choice there is will be a reduction in choice for that other OS and M$ will lose a bit more. It is hard to estimate the magnitude of this effect, but if Apple sells 3% of PCs and puts MacOSX on them then entrepreneurs trying anything to carve out more sales may easily flood the market with non-Apple PCs with MacOSX. Their prices will be lower and they should get more share. I imagine there could be a rapid 10% share drop for that other OS because many recognize MacOSX already but do not want to pay the Apple-tax. This will give them access to something they like at a reduced price. Psystar would not be doing this if they did not have buyers. The world of PC-makers is a lot bigger than Apple in potential volume.

Of course, the more people who see that there is an alternative to that other OS, the better and GNU/Linux will get some uptick, too. This compounding of effects will make 2009 a very good year of GNU/Linux. This matter could be settled as a matter of law, not requiring a trial as the judge can interpret the copyright laws for himself. I would not be surprised to get a ruling this year, perhaps not in time to accelerate the avalanche but that would surely come next year.

What of the merits of Psystar’s case? Copyright is supposed to be about allowing the creator of a work to profit by giving exclusive right to reproduce… Psystar can buy a copy to Apple’s benefit and install it on any machine they want. The copyright law does not permit a writer to dictate his book may only be laid on green felt tables. It permits him to receive a royalty, not obeisance.

This could be fun. This changes the whole universe as far as ultimate market share, post-monopoly. Assuming consumers are equally enamoured of similar products, one would expect a share of 1/N so we have 1/3 instead of 1/2, counting that other OS, MacOSX and GNU/Linux as the new three. When it was GNU/Linux and that other OS, 1/2 would have been about right. The difference between 1/2 and 1/3 is 1/6 of the universe of PCs, about 200 million machines. Multiplied by $100 or so and it is many billions of dollars. Real money and it flows out of M$’s clutches into Apple’s.

When OEMs see their share due to M$ fall, will they look elsewhere? You can bet on it. Some will switch to MacOSX or make it an option. Others will try to increase share by using lower prices for systems with GNU/Linux. Competion on price and performance. What a novel concept. One wonders why Apple does not embrace the future. I would bet they could sell enough copies of MacOSX to compensate for their loss of business on machines or they may actually be able to sell more Apple machines by cutting their prices… It is all good. Go Psystar.

- Robert Pogson

FUD Failed

You have read the FUD:

  • You will have to spend a lot to migrate to GNU/Linux
  • You will need some particular app not available on GNU/Linux
  • You will not save any money
  • TCO higher with GNU/Linux
  • etc.

Well, it is not true. You can do it. Here’s how the French police did it.

  • wake up to the reality of the Wintel treadmill
  • wake up to the reality of the cash flow to licensing
  • choose to use FLOSS apps
  • when all the FLOSS works and you use the browser for apps on the server, change the OS
  • because your users are used to the FLOSS apps, there is only a small cost to migrating the OS.

They saved millions of Euros over several years with this approach. In a few years 100000 employees will be free of that other OS completely.

This is a recipe that works, but I still like making the switch on a weekend. ;-)

- Robert Pogson

Patents on Software

Robe Enderle recently wrote, “If you drill down to the comment section of this earlier piece, you’ll see it filled up with folks arguing for and against software patents. While I’m in agreement that the U.S. patent system is currently far from perfect, I also agree with Chief Justice of the Supreme Court Rehnquist” that software can be patented under current law. And I believe, based on the foundation for the law, that software would have intentionally been included had it existed when the law was written. That position appears to be currently backed by the U.S. Supreme Court, but only by a 5/4 majority, which is hardly a ringing endorsement. So things could change. I’m just not aware of anything near-term that’s likely to get to the Court that would force such change.”

Sure, you can get a patent on software if it is part of a machine of some kind, but that is not what FAT32 long filenames are about. They are an information storage system, an abstraction of bits, not a machine. You cannot obtain a patent on long titles for books. You get a copyright on a particular title. If M$ thought its IP was violated in 2009 by TomTom, why didn’t they think it was violated in circa 1998 when GNU/Linux first used a FAT32 driver? There is a principle in law that you cannot allow a violation of rights to continue until a convenient moment to maximize damage claims.

The patent will not stand a good counter-attack by TomTom and the allegation of violation of patent is pretty weak in view of M$’s lack of litigation.

see http://www.uspto.gov/web/offices/pac/mpep/consolidated_rules.pdf

see http://www4.law.cornell.edu/uscode/35/

Whoever invents or discovers any new and useful process, machine, manufacture, or composition of matter, or any new and useful improvement thereof, may obtain a patent therefor, subject to the conditions and requirements of this title. “

That does not sound like FAT32.

(a) Except as otherwise provided in this title, whoever without authority makes, uses, offers to sell, or sells any patented invention, within the United States or imports into the United States any patented invention during the term of the patent therefor, infringes the patent.

TomTom did not sell FAT32. Why is M$ suing?

Patentees, and persons making, offering for sale, or selling within the United States any patented article for or under them, or importing any patented article into the United States, may give notice to the public that the same is patented, either by fixing thereon the word “patent” or the abbreviation “pat.”, together with the number of the patent, or when, from the character of the article, this can not be done, by fixing to it, or to the package wherein one or more of them is contained, a label containing a like notice. In the event of failure so to mark, no damages shall be recovered by the patentee in any action for infringement, except on proof that the infringer was notified of the infringement and continued to infringe thereafter, in which event damages may be recovered only for infringement occurring after such notice. Filing of an action for infringement shall constitute such notice.

Where has M$ marked a FAT32 filesystem with its mark? Oops. FAT32 is a specification, not a machine, so it cannot be so marked. No damages.

Except as otherwise provided by law, no recovery shall be had for any infringement committed more than six years prior to the filing of the complaint or counterclaim for infringement in the action.

M$ let go a big pile of loot, I guess. No, wait. They knew there was no violation and took no action. A jury will love that…

Looks like M$ has a pretty shaky case. If they wanted to protect their rights they should have done so ten years ago. They did not because they had no rights and they knew it. This action is FUD to frighten businesses into paying homage to M$. They are a bully. Do not buy their stuff.

UPDATE: TomTom and M$ have settled. TomTom paid an undisclosed amount and plans to quit using FAT32. I hope they paid 1 cent.

SJVN has a good article analyzing this development. see “Analysis: Microsoft-TomTom settlement is end of a battle, not the war

If everyone chucks FAT32 from GNU/Linux, GNU/Linux will no longer be able to shine with USB storage devices as it has. This is yet another means of M$ messing with the competition. I think we need to stand up to the bully and chuck the software patents. The SCOTUS should make some decision within the year.

USPTO:”However, based on Supreme Court precedent and recent Federal Circuit decisions, the Office’s guidance to examiners is that a § 101 process must (1) be tied to a particular machine or apparatus, or (2) transform a particular article to a different state or thing. This is called the “machine-or-transformation test”. If neither of these requirements is met by the body of the claim, the method is not a patent eligible process under § 101 and should be rejected as being directed to non-statutory subject matter. There are two corollaries to the machine-or-transformation test. A mere field-of-use limitation is generally insufficient to render an otherwise ineligible method claim patentable. This means the machine or transformation must impose meaningful limits on the method claim’s scope to pass this test. Also, insignificant extra-solution activity will not transform an unpatentable principle into a patentable process. This means reciting a specific machine or a particular transformation of a specific article in an insignificant step, such as data gathering or outputting, is not sufficient to pass the test.

Reading that, one cannot be persuaded that an algorithm like long file names is patentable as it does not transform anything real and it is not tied to a particular machine like a widget-maker. There is a good discussion of the fallout of the Bilski case here. It is very clear that FAT32 is not about any physical thing and is not patentable.

- Robert Pogson

Mother, May I?

When I was a kid we used to play a game which I have forgotten but one had to ask politely to make your play. I am reminded of this ancient game from an article published recently in Eweek.

Of course, M$ wants to suppress any technology that would lessen the number of fat clients running that other OS. What strikes me as outrageously funny are the lame arguments made:

  • company officials said Microsoft VDI might be suitable for implementations of up to 100 seats, but no more. Microsoft does partner with Citrix Systems for large-scale implementations that also include VDI.
  • “One of the key messages from the workshop: Virtual desktop infrastructure is an expensive proposition unsuited to large implementations.”

That’s like Canute commanding the sea to roll back. HAHAHAHA! <gasp> At least Canute used the commandment as a lesson. M$ thinks people will take its advice seriously.

From experience, we know:

  • even for one seat virtualization can be useful by keeping heat and noise and bulk out of the workplace, and
  • virtualization is an excellent way to keep a server working hard, earning its cost, and
  • virtualization scales very well for desktops because humans are rather slow peripherals, and few of us number-crunch. If we do, it is likely to be on a server or cluster somewhere anyway. Servers are better number-crunchers because they can have more power/RAM/CPU/storage.

An extreme example of scaling well is the GNU/Linux terminal server, even an old 32-bit server can run 30 users typing and clicking. A modern 64-bit job with many processors and multiple gigabit/s links can run a lot of users. In fact, if you match the size of the server to the number of users well, the cost of the server per user can be of the order of $25 dollars. That is scaling.

A couple of years ago, I had some dual-core terminal servers. Whether I had 20 or 40 users running on them they rarely got above 20% CPU usage and were never stressed. The terminal servers cost us about $1200 for hardware and ran Ubuntu so we paid about $40 per seat on the server. With today’s prices, $25 per seat is easily reached. We rarely had enough simultaneous users to try 60 users per server but it probably would have worked. Now we can have servers with 8 quad-core sockets and many gigabit/s links. Servers or clusters of servers for every conceivable workload exist. With shared memory, any ‘NIX can do a fine job of this, but M$ without shared memory tops out quickly.

It is interesting to hear M$ promoting GNU/Linux. Way to go, guys. ;-)

- Robert Pogson

Nightmare on the Ethernet

Thanks to M$’s incredibly fragile OS, Conficker is set to detonate on April 1. 12 million machines times, say, 1 megabit/s, is … 12000 gigabit/s, perhaps a giga-email/s.

All the security-through-obscurity/auto-updates/patches in the world cannot protect that other OS from malware and still people buy it. Perhaps this one will be bad enough that people change.

The way to fight malware is to keep stuff simple and clean so that the vulnerable interfaces are humanly countable and can be re-enforced. Instead, M$ adds feature after feature telling customers that they must buy another licence to use the newest features. The world is so tired of spam and malware that many are turning to GNU/Linux in 2009 partly to escape this stuff and partly because there is good software with good features in GNU/Linux.

Just like M$’s feature-ridden OS, Conficker is featureful malware. It has armor, unpredictability, diversity, counter-measures and now it has untold thousands of PCs waiting for April 1. Perhaps we will get some early warning of what it plans when clocks set wrong start to go early. Perhaps nothing at all will happen. Perhaps we will be without e-mail for a week until every IP running Conficker is blocked. Perhaps levels of spam will reach new records.

When the stuff happens, do not scrap your PC to buy a newer, more secure one. Chuck that other OS, not your machine. M$ and its partners laugh all the way to the bank when your PC quits working. Many take it in for service, meaning a re-installation of the same garbage that allowed the infections to take over. Many buy a new PC. Either way, the evil empire profits and nothing will change. You have to decide to end life on the Wintel treadmill. Break free. Enjoy your PC. Have it work for you, not against you. Use Free Software. It is designed to work, not fail at the drop of a hat. It is not designed to make someone rich at your expense.

Instead of a nightmare, have a pleasant and efficient experience with GNU/Linux. This year I have introduced hundreds of students to GNU/Linux. Every one was impressed by the speed and simplicity of the software. They are even more impressed to learn that I have never seen malware take over a PC running GNU/Linux in eight years on hundreds of machines. In that same period I have seen dozens of instances of malware even on patched systems because the malware is out there and operating long before the patch can be created.

Students are shocked to see the difference in speed between GNU/Linux and XP running on identical hardware side by side. They see factors of 3 in speed in favour of GNU/Linux. They see more than that when the machine boots as a thin client, obtaining a minimal OS from the server over the LAN and showing the pictures and receiving clicks while the server does the heavy work. Some of that difference in performance is the pathetic need to check every file for malware before opening it. That makes the filesystem perform like the world’s worst. I continue to be amazed that people are not aware of what malware costs them every time they use their PC. Perhaps Conficker will educate some more.

Is your PC patched against Conficker? Did the patch go in before the machine was infected? Sleep well tonight.

- Robert Pogson

2009 Could be the Year of Consolidation

There is news that IBM is in talks with SUN Microsystems about a sale. SUN is not a pure GNU/Linux play by any means but they are a huge part of the FLOSS community:

  • MySQL
  • OpenOffice
  • JAVA
  • OpenSolaris

to name a few big parts they play. IBM is a huge investor in FLOSS because it is what IBM’s customers want. Such a move could bolster FLOSS on servers and desktops and the cloud… It is all good and it is nothing to do with that other OS.

There have been many announcements of layoffs of employees in the economic downturn. Might we see some mergers to bring together compatible/complementary companies to provide diversification with fewer staff per unit of production? The longer the downturn, the more likely that is. Capital is scarce so firms needing new capabilities may absorb it rather than purchasing it. As long as this does not harm competition, the world can benefit from this if it keeps good companies going. SUN has had difficulty making money but they have made some brilliant developments in hardware and software. IBM knows how to sell/move stuff. It should work for those two.

On the FLOSS side of things, what mergers might make sense? There are fewer reasons to merge for economic reasons but cross-fertilization is another matter. We have seen the relationship between Edubuntu and Ubuntu change this year. Could we see distros with similar packaging cooperate more? Could Debian and Ubuntu get together somehow? The differences in personalities of the two organizations make that difficult even though they share a lot of code.

We have seen some mixed signals from Novell about how SUSE is doing. Could they be acquired by M$, for instance? As M$’s business model fails, something like that could make sense, but there is no way their stock-holders will let go of the old model any time soon. Could RedHat and SUSE merge? RedHat and SUSE are so different even though they may share very similar customers… Would the world be a better place with them working together instead of fighting? Would IBM look at RedHat if M$ acquired Novell? UnitedLinux fizzled. Could a new edition spring forth now that SCO is dead?

Hardware makers are suffering. I would expect some mergers in RAM/storage in this situation. We saw Seagate/Maxtor. Are there others that make sense? Probably the oversupply of RAM could re-shape the industry. How much RAM does the world need? A lot less if the world goes to the cloud or thin client/server computing. Perhaps some FABs will close because Moore’s Law means we do not need to plug as many sockets to get the job done. Vista tried to save them but even it could not.

Makers of motherboards are in for a change. The size of motherboards must drop the way Moore’s Law has been working. When was the last time, you saw expansion slots plugged? I made a few multi-seat thick clients with five video cards. Mini-ATX is good enough for most people. That means more thin clients/netbooks and less tonnage of e-waste, not a bad thing but it does not take as many people and as large a facility to work with such tiny equipment. Everyone has to change. Change is good.

What about USB 3? Does that change everything? It could. Will investment in the new technology be made in the downturn? Yes. Everyone will fight for their share of the shrinking market. Consumers will win. Businesses that held on to old technology in the face of Vista will win. When they do upgrade, it will be to vastly superior stuff and probably with a lot more GNU/Linux. It is all good.

- Robert Pogson

2009 is the Year of GNU/Linux

Seriously, GNU/Linux will make great progress this year while that other OS languishes in doubt about vapourware and a current product with tarnished reputation.

IDC recently did a survey sponsored by Novell. Two key questions of a sample of the world’s small and larger IT organizations:

Q. Do you plan to increase your adoption of Linux on servers in 2009 as a result of the
economic climate and a focus on cost containment?
Q. Do you plan to increase your adoption of Linux on clients in 2009 as a result of the
economic climate and a focus on cost containment?

The answer was “yes” in about 50% of respondents, “maybe” in 20% studying the matter and “no” in only 30% of respondents. Of course, the degree of adoption is still unknown but it is costly to maintain support of a newer system if it is only a small part of the organization, so I presume these are significant migrations.

In my organization, the result is the same and we will move about one-third of seats to GNU/Linux. That would have been done already except we have done some work to get common authorization working for GNU/Linux and that other OS. We could easily replace that other OS for all the student-seats but there are conservative elements about… ;-) . This use of AD permits accessing files shared with AD security. Thanks,
WinBind, and Debian. The terminal server added to run the new clients will provide an abundance of storage so we could simply move all the students data and make Samba the domain controller, freeing up a powerful server to run the rest of the system. It all works, except spaces in usernames. We can switch to XFCE instead of GNOME to dodge that problem. or we could take out the cursed spaces.


- Robert Pogson

Big Bang!

IDC has a forecast of PC production for the next few years.

Highlights:

  • flat or downward trend for 2009.
  • recovery in the world excluding USA in 2010
  • recovery in USA in 2011
  • significant pop as commercial equipment renewal kicks in

This gives FLOSS another year to take advantage of M$’s weakness. If, after 2010 many are going to bite the bullet and invest in new PCs of all kinds, what OS are they going to choose? The one that costs more to maintain fighting malware/DRM/bloat or the lean/mean computing system GNU/Linux? We know from the KACE survey that many have looked at FLOSS. Two more years of looking should be useful. Thin clients are hardly slowing. Netbooks are hardly slowing. Both should be seen as viable technology by the time of recovery and GNU/Linux works well with them.

Two more years of Moore’s Law will give us thin clients well under $100 and shares of servers that will run a user for $10 or so. Retailers are madly trying to sell larger monitors, fancier mice and more powerful video cards, but enough is enough. We can do what we need with today’s technology and someone will supply it even if the Wintel monopoly keeps wanting us to upgrade in an ever more uselessly tight spiral. Something has to give. It is monopoly. This is the end of it.

In the downturn all the little guys of IT will struggle to find a niche which has growth: netbooks/FLOSS. In two years they will be able to supply much of the world’s need for IT without Wintel. Wintel will have to cut prices. We do not need many more features or more performance. Both hardware and software are now commodities that no small cabal controls. The cash cows have dried up and productivity and efficiency will yield profit, not monopoly. Expect continued strong growth in EMEA/BRIC/APEJ. The USA and all who cling will be seen as foolish to throw money at Wintel, just as they are for driving Cadillacs in a world with tight supplies of oil. Furthermore, a considerable portion of spending on IT in the recovery will be on cloud-computing, not pumping money into the monopoly but services off-campus.

“Over the next four years, IDC expects spending on IT cloud services to grow almost threefold, reaching $42 billion by 2012. More importantly, spending on cloud computing will accelerate throughout the forecast period, capturing 25% of IT spending growth in 2012 and nearly a third of growth the following year.”

While the downturn and the recovery will be modest by the usual measures of the monopoly, this will be a big bang of unprecedented proportions for FLOSS and hardware. Instead of more cores at higher clockspeed, a few cores at lower clockspeeds will yield the ultimate thin clients, tiny modules inside monitors or keyboards. Instead of pricing that the market will bear, money will be made on high-volume, low-cost parts that can be made anywhere in the world.

  • One or two servers will be enough for SMBs like schools. These may even be off-campus in the cloud or head-office.
  • No need for licensing fees or maintenance contracts because there is so little to go wrong.
  • Scarcely any need for in-house maintenance.

Then, we can get on with our job of finding, changing, creating and displaying information instead of propping up monopolies. Freedom at last.

- Robert Pogson

The Other Half of Wintel

I spend a lot of text on M$, but there is another part of the Wintel monopoly: Intel. Intel is the company that brought us inexpensive microprocessors in bulk. I remember the joy of reading about the 4004 chip in Electronics back in the 1970s. That was the beginning of Moore’s Law and the incredible process bringing powerful computers to the masses of humnanity not in the Fortune 500.

Unlike M$, Intel had to fight for its monopoly. It produced a great product in competition with others entitled to the technology Intel had developed to ensure multiple sources of CPUs according to the deal with IBM, the developer of the IBM-compatible PC. Along the way, they became greedy and tried to take it all, excluding other chip-makers from the market. They became able to charge large price differentials between their chips and those of the competition. About the time “Intel Inside” became a household word, I was building my own PCs. Unlike others, I looked at the specifications and could appreciate that by buying AMD, I could achieve what I wanted at a lower price, sometimes $100 or greater difference. That changed when AMD produced the AMD64 line. For a time, they could charge $1000 for a new chip because Intel was not in the ball-game.

Now, Intel is a step ahead in Moore’s Law and Intel has CPUs that will run Vista with huge caches, but AMD is still worth supporting if only to keep Intel honest.

There is news of the details of Intel’s technique for excluding competition from the market in a translation of a legal decision in Korea:

“As Samsung, the No.1 domestic PC maker and the 100% defendants-CPU using company at that time, had been starting to buy AMD CPUs and launched AMD CPUs- installed PCs from the first quarter of 2002, the defendants continuously requested Samsung to cease from buying AMD CPUs. As Samsung rejected these requests, the defendants significantly reduced the volume of rebates in the first quarter and second quarter of 2002. Since then, the defendants continuously had requested Samsung to cease from buying AMD CPUs, exploiting the provision of rebates as an instrument. In May 2002, the defendants suggested Samsung the “long term support plan” with promise to offer maximum-level rebates from the latter half of 2002 and afterwards, on the condition of Samsung’s suspending the purchase of AMD CPUs by the third quarter of 2002, and, consequently, Samsung accepted this proposal. Since then, from third quarter 2002 to second quarter 2005, the defendants established the quarterly support plans pursuant to the “long term support plan,” and provided rebates to Samsung during the period. In accordance with these rebates, Samsung stopped purchasing AMD CPUs in the third quarter of 2002, and from that time, Samsung maintained the Intel CPU Market Segment Share (henceforth, `MSS’) at 100% level until the second quarter of 2005.” (17p)

“Reviewing this argument, as the Table 31 illustrates, considering the amount of rebates actually provided to Samsung and Sambo, we could observe that the rebate rate was in the lowest level in first quarter 2002 when Samsung rejected the defendants’ request for abandonment of AMD’s products, despite Samsung’s purchase amount of the defendants’ products was in the highest level at the same period. On the contrary, we could observe that the rebate rate was at the highest level in third quarter 2002 when Samsung stopped purchasing the AMD’s products, despite Samsung’s purchased amount of the defendants’ products was at the lowest level during the same period. This fact resulted from the fact that the defendants’ rebates had not been provided in pursuant to the amount of partner’s purchased volume but pursuant to the purchased volume of competitor AMD’s products. Accordingly, it is appropriate to conclude that the defendants’ rebate program is totally different from the volume discount scheme which defendants are arguing.”

see the translation by The American Anti-Trust Institute.

The reasons Intel did these things was because AMD had superior chips for a time and AMD had better prices. Intel could have lowered prices to improve price/performance, but they chose to illegally exclude AMD from the market. AMD is still doing well on the low-end desktops but they have nothing in the netbook game and Intel clearly has a lead in power/price/performance at the high end. In my game, GNU/Linux terminal servers, those huge caches are just the ticket for a machine running 1000 or more processes and making many context switches per millisecond.

I can assume Intel carried out such tactics globally. It is unlikely a rogue element in Intel did it only in Korea. In that case AMD v Intel may have legs. Let us hope the consumer gets a break in this and not just corporations. Otherwise, they will take our money, with the only doubt being who gets what share.

Competition in hardware and software ensures constant improvement and efficiency. Wintel is coming to an end thanks to much effort globally. Wintel has ripped the world off for billions but we do not have to take it any longer. Buy AMD, Via, ARM etc. and use Free Software. It is in your best interests to do so, even if there are short-term costs. The alternative is long-term over-charging and lack of choice.

- Robert Pogson

Migrating a Computer Lab From XP to GNU/Linux Magically

You don’t believe in magic in the 21st century? Try this:

  • take twenty-four eight to ten years old PCs running XP Pro taking minutes to boot
  • collect their network connections to one switch with two gigabit/s ports
  • add a GNU/Linux terminal server to the uplink
  • boot from a floppy each client, loading software from the server into RAM
  • run a script on the server to wipe the hard drives and install a boot-loader on the hard drives

Then, for a couple of hours of work, the client machines provide virtual desktops from a newer server and the users feel the speed and responsiveness of the newer server instead of the ancient PCs.

Perhaps it is not magic but only sufficiently advanced technology, but it is a wonderful way to upgrade a lab with very little equipment costs. The ancient PCs can be made useful again and give a few more years of service. When they finally must be replaced they can be replaced with new thin clients which cost $100 or less and are tiny, cool and quiet. This is a great solution for schools. Replace the management of many PCs with the effort required to manage one.

A few details:

  • The terminal server is an AMD64 3000 with 2gB RAM dating from 2004, not young but a generation or two more advanced than the clients. Most importantly, it has a 200gB RAID 1 array to speed disc storage. It also provides local services like chat/web/database so users have much better response than the Internet.
  • The clients work in an NFS share from the server which contains an LTSP 4.2 environment. Clients have two sessions, GUI and local BASH shell but also have keyed SSH access, useful for the scripting.
  • Script run on the server:

for f in list-of-ip-addresses-spaced ; ssh $f /nfsroot/script.sh& done

  • Script the client runs:

script.sh at /opt/ltsp-4.2/i386

modprobe ide-generic # loads driver

dd if=/dev/zero of=/dev/hda # wipes drive to free licence

dd if=/nfsroot/boot-loader-from-rom-o-matic.net of=/dev/hda #write netboot loader

haltsys

It cannot be emphasized too much how great an improvement this small effort gives to each user of the lab every day in every way:

  • faster booting because there is less software to boot
  • faster login because the first user to login loads all the necessary files into shared memory
  • faster loading of applications because the files are already in shared memory on the server, much faster than reading from a hard drive
  • files are accessed by processes on the server from drives on the server so there is no network traffic/bottlenecks for data
  • no need to run AV scanning for each user
  • web applications and data from the local server are much faster than going through the Internet bottleneck
  • RAID storage is much faster and more reliable than the hard drives on the client

I demonstrated this technology to teachers at a conference last week and they were amazed as one should be by magic.

Don’t have a GNU/Linux terminal server? Install Debian GNU/Linux or Ubuntu or K12LTSP. Debian packages to add is ltsp-server-standalone. It sets up DHCP for the subnet on the switch and the tftpd service to boot clients and sets up the file share. Every school should have a few of these terminal servers.

- Robert Pogson

Convulsions

Addicts who overdose on drugs often die in convulsions brought on by the effects on the brain.

A PC maker in the USA has decided to help out M$’s customers who buy high-end PCs with Vista to upgrade to 7, That is decent of them. The business eats $200 or so and the end-user pays the freight. That is a high cost for maintaining the addiction. It is good PR and stimulates business in a down economy but the end-user could:

Those all look like better options to me. Break the chains that bind you to your abuser. Avoid convulsions/contortions induced by the puppet masters. Be free. Use Free Software.

- Robert Pogson



Archives by Month

My Mission

My observations and opinions about IT are based on 40 years of use in science and technology and lately, in education. I like IT that is fast, cost-effective and reliable. I do not care whether my solution is the same as yours. I like to think for myself.

My first use of GNU/Linux in 2001 was so remarkably better than what I had been using, I feel it is important work to share GNU/Linux with the world. I have been blessed by working in schools where students and school systems have benefited by good, modular software easily installed in most systems.

I have shown GNU/Linux to thousands of students and hundreds of teachers over the years and will continue in some way doing that until I die in spite of the opposition.

Posts

March 2009
S M T W T F S
« Feb   Apr »
1234567
891011121314
15161718192021
22232425262728
293031