Billosaur wrote:"most CEOs are not in the habit of overturning the judgments of their technical people -- they are interested in the big picture of the company as a whole, not the minutia of day-to-day operations. The don't care what hardware/software is used, as long as it runs, works, and they don't have to hear about it from shareholders."
Some CEOs are responsible to a board elected by shareholders who want the business to make as much money as legally possible. If IT is a large share of the overhead and profits are marginal, there should be a market force pushing the business towards lower cost, such as GNU/Linux. We have seen that for servers, but not as much on the desktop. The difference is standardization. HTTP is fairly standard and a server pushing that will work whether it is Linux or M$ stuff. The desktop user has to be sold, comfortable and proficient with the user interface to a GUI. Whether it is true or not, many PC users believe they can use M$ stuff and other GUIs will be difficult to use. That opposes market forces based on price/performance of the software. Another thing is that many businesses are locked-in to particular applications that have been written for the M$ API (which M$ constantly changes to force upgrades) and it actually costs a lot more to change the application set than the OS. Some firms insist on spending thousands of dollars per employee just to re-train for a new OS and applications. For a few specialties, that may be effective but for many, it is a waste and used mostly to add to the barrier against change.
Where the opposition to change from the M$ way fails is that it just looks at the next couple of quarters. After the transition, the change keeps benefitting the balance sheet by continued improved return on investment. Many firms that migrate find break-even in one year, especially if they go from M$ on thick clients to Linux on thin clients. Often they can continue to re-use equipment that M$ declares obsolete. For thousands of seats, that is millions of dollars saved instantly. The cost of adding a powerful terminal server is similar to the cost of adding RAM to a bunch of clients and the labour of going around to each machine may be saved, except for a BIOS setting for network booting. Unfortunately, many PHBs say they have a system that works and if it isn't broken, don't fix it. They do not understand that their system is broken and never works as well as it could.
M$: It's time to pay us again what you owe us for that defective OS we sold you.
victim: We can't afford it. Widows and orphans need the money...
M$: What can you afford?
victim: $250000!
M$: OK, for that we will only break one knee. Remember this next time and don't be late!
victim: Oh! Thank you, M$!
---------
Every time I hear people willingly paying the M$ tax it makes me sad/angry. There are hundreds of millions of folks around the world who have just upgraded to XP, which was obsolete in 2001 when it was released, IMHO. I work in education where PHBs boast about being Wintel shops and there are classrooms with 0 or 1 PC in a classroom running that obsolete OS. It is all they can do to maintain a few labs where kids are scheduled to visit. If they used FLOSS, they could have twice as many PCs in schools and more peripherals for the same or less money. IT is not fulfilling its promise to education simply because of the M$ tax.
1arkhaine wrote:"It's a pretty sad outlook you have on a battlefield when simply being on it is apparently cause for victory. Don't you need to win, before you can win?"
M$ is a monopoly. Before you can win the war, you have to win the battle of getting into the market. Dell et al are helping but they are not alone, just the biggest and most recent.
see LXer pre-installed Linux database
It is well known that FLOSS has fewer bugs per 1000 lines of source code. The bloat that went into Vista brought in plenty of bugs to be sure. Key differences between Linux and M$ stuff:
M$ gets stuff determined by the sales department. We know how well salesmen design systems.
Linux is designed to be modular so the complexity of each piece is less. M$ has stuff where the browser installs code, printing a document can cause pieces of the file to be executed, etc.
There are far more projects in FLOSS than there are coders in M$. More manpower, with properly filtered output results in more correct code.
If a bug bugs me, I can look at the code, file a bug report, or suggest a patch. There is no way that can be done with M$'s way of doing things. Vista release was as buggy as a Linux release candidate.
"There is value in using computers for education in K-12."
AI and quizzing are just small parts of the value of computers in education.
saving paper by reading from the screen and moving document files instead of paper
course management systems like Moodle help plan, organize groups and speed up evaluation/feedback
searching full text using a local search engine or Google is thousands of times faster than flipping pages in books and more relevant to most work places these days. With swish-e one can have one's own search engine on every server and PC in our schools.
writing/editing/refining is much faster on a PC than with pen and paper. This also saves much paper if each draft is not printed.
a server can hold many thousands of texts that students and teachers can access from their desktops in a second. This is much faster and more efficient than having folks walk to the library and shuffle books and paper around. Every school library should have online access at least from within the school. See koha.org , for instance.
if a teacher has a class of 25 students for one hour the teacher can lecture for an hour and put them to sleep repeatedly or the teacher can give a short, sharp introduction and turn students loose with familiar computing tools to read, analyze and respond to media and text on the computer. Many students are much more productive responding privately to an interactive application rather than waiting in the queue for their 4 minutes of teacher time. We should not be teaching students to wait but to think and do.
Many school systems are concerned about the cost of computers but with FLOSS and older/lower-powered client machines, schools can run powerful Linux terminal servers for less than half the cost of that other OS on a fat client. With donated PCs wiped of that other OS and booting via PXE/LTSP, a school only needs a good network, powerful servers and a bit of care to set up the system for the different way of operating.
To colonize space one needs only package up the DNA, freeze it and ship it. DNA tinkers out there will re-assemble the humans for their lab experiments. If you suspect there are no DNA tinkers out there, just ship a robotic lab along. No need for life support along the way... just refrigeration and uninterruptible power supplies.
AMEN! I installed an LTSP system in a new school last year and had zero problems with Linux as a terminal server. The thin clients were quiet and trouble free. We put most of the money in the server room and obtained great value for the money. We had plenty left over to bury the school in printers, cameras and scanners. Students had no problem with it except that the BOFH could delete their processes if they were off task in the computer lab. Teachers had no problems except the one who needed hand-holding with everything. Linux was not his problem.
We had twice as many seats installed this way as we could have afforded with that other OS and almost nothing to do for maintenance except backup and watching it hum.
What those who compare LTSP and such with dumb terminals of 20 years ago is that Moore's Law had dragged the CPU, RAM, networking and motherboards into the 21st century. A modern CPU is idling on most desktops (80%) and RAM is cheap and plentiful on 64bit servers. LTSP is a compromise only for full screen video and heavy number crunching. Even then, a power user gets a slice of monstrous power on a multi-CPU Optern server loaded with RAM instead of a typical desktop. The school's load is almost ideal for LTSP except if you teach multimedia production in a lab. Those 20% of users who need it can still have a Linux thick client but that solution is unnecessarily costly for every seat.
A similar technology that multiplies benefit/cost is the multi-seat X client. You can hand 6 to 12 users on one PC using single or dual head video cards and multiple keyboards and mice. Groovix and Userful are two implementations. Groovix is open source.
Unless you can describe something in numbers your knowledge is of an uncertain kind.
There is reason to believe that there are 1000 million PCs in the world and 200 million new ones are being produced each year. Even if folks were willing to switch to Vista, it would take five years and nearly a trillion dollars would be spent on this Vista foolishness. That might make sense if there were a real benefit to Vista. Consider the alternative scenario. Instead of junking 200 million PCs each year, migrate them to GNU/Linux and they keep running. This migration will either quickly double the number of PCs running in the world or slow down the migration to Vista as being unnecessary. This is not like going from a Pentium I to a Pentium III or some such upgrade. Current PCs are mostly Pentium III and better, working very well. It makes more sense to migrate working PCs to Linux than to chuck them and go to Vista with unnecessary hardware.
People know their PCs are good enough because of what they can do with XP. They will not be fooled. When XP is no longer supported in 18 months, the world will be overrun with malware as the XP machines stay on-line. Folks will have to migrate to GNU/Linux.
why is it that when ever MS EOL's a product there's mass outrage
The world and Microsoft have spent billions getting the stuff to work. Then Microsoft kills it and the world has to buy new equipment, pay Microsoft more money and start all over again. There would be no objection to Microsoft continuing to improve a product that many find useful and there would be no objection if hardware could be continued in use for its working life instead of the life of Microsoft's product. The reason that Microsoft cannot continually improve its product is that it is spaghetti code (not modular) and everything changed has unfortunate unintended consequences affecting other parts. That particularly ticks off developers who know how code should be developed. On top of the life cycle issues, the poor design and large changes to the new product introduce a new set of vulnerabilities that drive admins batty.
Look at the recent release of Debian 4.0 Etch. It could be a large jump from Sarge, but it is not because Sarge users could follow the developments in testing, watch bug reports in Etch and make the change when it suits them. OSS is just a better process. Watching hundreds of millions of PC users go down the wrong path is like watching genocide. Good people just have to say and do something about it.
All true, but many want to use their systems for productive work, not debug them. If you develop software and want to live on the cutting edge, testing/sid is your environment. I teach. There is nothing worse than showing an interested crowd some new software and having something freeze. Ask Bill Gates: http://edition.cnn.com/TECH/computing/9804/20/gate s.comdex//
Last year I installed a new system based on LTSP (http://ltsp.org/) on EdUbuntu. It was a few days before school started and I had the staff in my computer lab to show them the basics of the new system (teachers are usually not adventuresome like kids...). I was a few clicks into the presentation when a display driver started corrupting data eating away at the bottom of the screen. It turned out that good old Vesa driver did the job, but I did not figure that out during the presentation. What a flop I was! My boss dozed off! When school opened everything went smoothly except I did not have LDAP properly configured so some users had different passwords on different servers (not Debian's fault) and some memory going bad in one server. There are times when the software just has to work and that is much more likely with something that has existed in testing for quite some time.
etch has been less buggy than sarge for almost a month.:)... and less buggy than that other OS since Etch began to boot. It is absolutely amazing that a loose, democratic organization like Debian can out-perform a giant of industry with giga-bucks in the bank and tens of thousands of slaves but it is true, the bazaar works. Freedom and empowerment take some work and spine but the result is worth everything many times over.
Debian is one of the great old distros that just keeps getting better and not by adding frills. It is a large distro on many architectures supported by package managers from around the world. It is not hard to install as the reputation was. It is huge with many thousands of packages all smoothly (well, mostly;-) integrated. I favour it for anyone migrating from that other OS, a new installation or on a large or small system.
One of the neat features of Debian Etch is the smooth set of packages for installing LTSP (See http://ltsp.org/ ). One can go into a school on the weekend, set up a server and support all the old equipment as thin clients whether they be iMacs, i386, i486, P-what-evers and manage hundreds of accounts by Monday.
I have been using Testing for a couple of months and there are few bugs. Nothing has prevented me from using it in production.
Customers are demanding service from hardware manufacturers for Vista drivers, not from Microsoft. HP and others want to sell new hardware not patch old drivers. The customer wants his working device to continue working.
The last gasp was 2001 when they brought out XP. It took them five years to fix its bugs and now they have to kill XP in order for Vista to have any chance. That last gasp brought in some oxygen to keep the beast barely alive, but it is all gone now. If they had done an Apple and put some UNIX underneath they would not have to be a bully to protect their turf. They would not need to throw out silly features to sell the stuff. It would be good. MSFT made it to monopoly by being in the right place at the right time and it would have kept monopoly if the product had been any good. Instead of improving it by using sound design, they kept on adding crap and using dirty tricks to keep the monopoly. You can only fool all of the people for a period of time and they wake up eventually.
It is too late now. They have burnt too many bridges. Even the hardware makers hate MSFT because they changed VISTA just enough to break all the drivers. Millions of school kids are experiencing the richness of GNU/Linux. Dell and HP are getting serious about Linux machines (maybe). The EU may realize that mega-fines are not enough and outright ban MSFT. Even Uncle Sam is discouraging the use of Windows for security. MSFT is like a long-necked dinosaur trapped in a tar pit. The brain is so small and so far away that it has not received the final BSOD yet. The body is cooling slowly because of the large mass to surface ratio.
"Microsoft has to learn to deal with the fact that they have to compete and can't release any old rubbish."
A flat quarter with a new product that cost billions to develop should do it...
Not that I think MSFT will learn to be more competitive, and they have burned too many customers, offended too many governments, produced too much schlock, told too many lies and cost the world economy too much to be welcome in the market place.
I am in a place where the curriculum is from 1992 (the last millenium...). There seems to be a reluctance to mandate PCs in every classroom because of the costs.
I strongly advocate that every geek drop in at a local school or school division and talk to them about IT. The existing clunkers may make fine thin clients and a few new servers do not cost much to use as Linux terminal servers (see http://ltsp.org/ ). There are several distros that automate the conversion of a reasonable PC with two NICs, some extra RAM and storage into a Linux terminal server so that the old machines or new thin clients have only to show the pictures and receive the clicks. Thus a whole lab gets to log in and run on a single good server. Maintenance is down an order of magnitude this way as only one machine needs a file system. Debian, Ubuntu, SkoleLinux and K12LTSP have reasonable repositories for schools. These system do everything a school may need except allow full screen video to every seat. All the normal click/gawk/click stuff works beautifully. Teachers can easily monitor/control each student using VNC or whatever. I like to log a student out when they wander... or I block their favourite time-wasting sites. These are powerfull payoffs for schools who make a small investment in effort to install such a system. It is useful in a single classroom, a lab, or a whole building.
Schools that have a bit of money to spend can invest in a powerful server that can run a whole school over gigabit/s. A motherboards like TYAN S3992 (see http://www.tyan.com/product_board_detail.aspx?pid= 235 ) are just made for this with Lots of RAM, dual gigabit/s NICs, and dual Socket F Opterons. The investment in such a server is spread over every seat in the system and the per-seat cost can be $25 or so for the server, free systems donated by government or business ($0) or new thin clients such as NTAVO 6040 ($139) with LCD screen, USB keyboard and mouse really is very cost effective and easy to maintain. Having a single server is a little less reliable as a single point of failure but having fewer parts also saves money.
I have found Windows to be exceptionally secure after I install GNU/Linux right over top of it. I have never had a failure after this procedure. I started doing it when I saw machines running Windows fail for no apparent reason, sometimes just idling and "PFFT!", dead as a BSOD. People told me it was hardware problems, but, running on the same hardware, the new installation would run for months with no downtime.
"Hell, getting our IT to respond quicker than a three-week time period to a simple "Please reset my password" request would be nice."
Sounds like "Inactive Directory" at work. On my last system, after a few requests from folks who needed to change their passwords but did not know how, I typed a single line command that put an "apple" icon on every desktop with the words "change my password" underneath. That change went out to four servers on encrypted connections and it was so, just like God at Creation. Those who came to me in person had their passwords reset and they sat down at a chair next to my terminal and logged into their own configuration and changed their password to suit them. With folks in a WAN, that would not be possible and there would have to be some levels of filtering to prevent hackers from phoning up and asking for Joe's password to be reset... I created 700 accounts in a few minutes on my system with random passwords and usernames and handed books of coupons to managers who issued them to underlings and recorded the transactions. The whole user system was set up in an hour at my end. I detected two of those accounts being misused and suspended privileges instantly. No sweat.
Linux is the right way to go. With simple scripting one admin can control/monitor hundreds of servers with similar configurations. Of course, he can futz it up pretty quickly, too, but that is why Linux sysadmins draw top dollar.
Exactly. With Linux, you set it up and it just runs. There are none of the problems you encounter frequently with that other OS. Linux is the key. It works for us not Bill G.
I switched to Linux when I had just five Windows machines that crashed daily just when someone needed them to perform. I put Linux on out of desperation and the same hardware ran six months without downtime. People who use that other OS just cannot get their minds around a system that works, has few bugs, costs little, and is flexible enough to do whatever we wish.
Many cannot imagine watching a programme run on one machine from another machine. In the Windows world, that is two points of failure. In Linux, it is the way the display system works. You can have as many displays as you have resources and they do not have to be on the same machine. Bandwidth is only an issue if everyone is watching movies instead of working...
First warning... end users that these terminals are targeted towards typically HATE having to use a terminal like this where the software is served remotely. Unless if you have enough server bandwidth and they are local enough to be able to deliver the needed software at a speed similar to using it locally on a PC, you're going to be doing nothing but frustrating the end users. For people that really need to do the work, they want their software to run as quickly as possible so they can get their job done as quickly as possible. Running it remotely is only going to slow things down.
If it is done right, the user of a thin client does not even know his programme is running on a distant server. Normally, a user will have something like a P4 on his thick client and a single hard drive. When he loads a file, it takes a few ms to seek and a few more to transfer. With server-centric computing, files like the browser executable, common webpages, and parts of the databases are already in RAM taking microseconds to activate. Things just flash on the screen! On my personal terminal server, now two years old and rickety, the first time I load OpenOffice (my biggest, ugliest app) it can take 7s. When someone logs in after me and runs OO, it takes 2s to get going. You tell me users hate that?
The truth is the typical thick client is idling at a small percentage of CPU load waiting for a mere human to read and click. The idea of server centric computing is to give such unused power to the next guy. My AMD64 3000 (1.8 gHz clock) can please a whole room full of people (I have run 30 simultaneous users) on this basis alone. If you add to this caching of files, huge buffers, and RAID storage, you should ask, "Why would anyone want a pokey thick client that needs software and hardware maintenance and replacement every few years?" My server cost $1200 to build (about $50/client). Programme loading and file serving uses no bandwidth. I have everything on the server. I have gigabit/s bandwidth and 30 megabits/s is plenty for one client.
For 3500 clients, you could install 120 servers like mine of 30 quad core machines or 20 quad machines with faster clocks. The multi-opterons do cost more but you can put way more than 4gB RAM on them.
In case anyone is wondering, I run Linux, not stuff designed to run a single user.
I worked at a place one year where MSFT decided to override our settings to "upgrade" to SP2 and our whole system needed reinstallation. We did not have enough storage to back up every client so I had to go through the whole system customizing clients. It made my day. In systems I design there will be no Windows.
The last system I designed had 130 seats as Linux thin clients and I could tweak the whole system without leaving my chair in seconds. I had redundant servers ($1500 each) instead of redundant clients and it took only minutes a day to verify that everything was OK and it was for months. Not one incident of malware disrupting anything. The users migrated from needing a full time geek to re-install that other OS several times each year on each client to having machines as reliable as telephones.
Yes, XP/2003 does this kind of thing. A power user like me with gigabytes of files does not enjoy moving around in such a system. Synchronizing...
Give me LTSP or even a remote X session any day. My files stay on the server. Only pictures of their contents and my desktop follow me around. Makes a lot more sense.
Moore's law increases the power you can put in boxes in the server room so rapidly that you do not even have to increase the size of the server room. Just keep upgrading the servers. A few years ago we jumped to AMD64. Computing MFLOPS did not increase much but throughput sure did. Now we have higher clock speeds and multicore chips. Server-centric computing just makes sense. There is no need to upgrade the thin clients for many years and yet we can constantly redouble our computing power in the server room. So what if we have to water cool them? I like to keep all that noise and heat away from my desk.
I disagree strongly with this statement:"The cost savings of opting for a truly non-capable display-only device over a competent computer is essentially non-existent."
With Windows it is true, because Bill charges per-seat. With Linux you get about 66% discount per seat. The cheap thin client box may cost only $150 USD. You pay Bill more than that in licences.
I recently equipped a whole school with thin clients in every room and we got twices as many seats as we could have afforded with that other OS and the cost of maintaining the thin client is almost zero. Over ten years, they will save their own cost in power savings, so they cost almost nothing. On the other hand, a thick client costs over and over again and is a burden rather than a blessing. There are a few things a thick client can do that a thin client cannot, but that can be worked around: go thick for the few individuals who need thick. Why pay big bucks to have thick clients idling all over the building? Pay big bucks for servers working hard.
"Again, businesses don't want anything that deviates from tried, tested and true path. Cost is not a concern here; labor and apps cost uncountably more than the OS."
If businesses A and B have 100 seat computer systems and A pays tens of thousands of dollars for software and B uses FLOSS, B has an advantage. For seats that involve the usual office stuff (word processing, spreadsheet, image processing), GNU/Linux works today on the desktop.
I think most of the "special" apps with the high price tags are archaic and represent lock-in. What do they do that requires a particular piece of software and no other? I have heard many say that they can work with PhotoShop but not with Gimp and I ask for details and get some tiny detail that is an insignificant part of the task.
I was once involved in a project that was locked-in with software that took six man-years to develop. The developer went bust. I replaced the software with all the functionality we needed in six man-weeks of writing from documentation and the result was better. We used to train employees to use the old programme (and the industrial systems it controlled) in six months. The new programme took only one month of training because the user interface was greatly improved (hints at the bottom of each screen). The entire cost of the transition was recovered in the first year because it took so little time to train users. This was mission-critical software without which the system could not run. The cost of the whole system was millions of dollars and it was in jeopardy if we needed to change anything in the software, so we replaced it with something we controlled. It made business sense, was cheaper, and it worked better.
I recently moved a school from Windows to Linux with no downside. Everything works. We control our destiny. We saved a ton of money on each seat and could afford twice as many seats as was planned when Windows was the concept. Linux is ready for the desktop certainly for many users who browse, write, move and store stuff.
Slashdot Mirror
Some CEOs are responsible to a board elected by shareholders who want the business to make as much money as legally possible. If IT is a large share of the overhead and profits are marginal, there should be a market force pushing the business towards lower cost, such as GNU/Linux. We have seen that for servers, but not as much on the desktop. The difference is standardization. HTTP is fairly standard and a server pushing that will work whether it is Linux or M$ stuff. The desktop user has to be sold, comfortable and proficient with the user interface to a GUI. Whether it is true or not, many PC users believe they can use M$ stuff and other GUIs will be difficult to use. That opposes market forces based on price/performance of the software. Another thing is that many businesses are locked-in to particular applications that have been written for the M$ API (which M$ constantly changes to force upgrades) and it actually costs a lot more to change the application set than the OS. Some firms insist on spending thousands of dollars per employee just to re-train for a new OS and applications. For a few specialties, that may be effective but for many, it is a waste and used mostly to add to the barrier against change.
Where the opposition to change from the M$ way fails is that it just looks at the next couple of quarters. After the transition, the change keeps benefitting the balance sheet by continued improved return on investment. Many firms that migrate find break-even in one year, especially if they go from M$ on thick clients to Linux on thin clients. Often they can continue to re-use equipment that M$ declares obsolete. For thousands of seats, that is millions of dollars saved instantly. The cost of adding a powerful terminal server is similar to the cost of adding RAM to a bunch of clients and the labour of going around to each machine may be saved, except for a BIOS setting for network booting. Unfortunately, many PHBs say they have a system that works and if it isn't broken, don't fix it. They do not understand that their system is broken and never works as well as it could.
M$: It's time to pay us again what you owe us for that defective OS we sold you.
victim: We can't afford it. Widows and orphans need the money...
M$: What can you afford?
victim: $250000!
M$: OK, for that we will only break one knee. Remember this next time and don't be late!
victim: Oh! Thank you, M$!
---------
Every time I hear people willingly paying the M$ tax it makes me sad/angry. There are hundreds of millions of folks around the world who have just upgraded to XP, which was obsolete in 2001 when it was released, IMHO. I work in education where PHBs boast about being Wintel shops and there are classrooms with 0 or 1 PC in a classroom running that obsolete OS. It is all they can do to maintain a few labs where kids are scheduled to visit. If they used FLOSS, they could have twice as many PCs in schools and more peripherals for the same or less money. IT is not fulfilling its promise to education simply because of the M$ tax.
M$ is a monopoly. Before you can win the war, you have to win the battle of getting into the market. Dell et al are helping but they are not alone, just the biggest and most recent. see LXer pre-installed Linux database
see Cyberinsecurity at http://www.ccianet.org/filings/cybersecurity/cyber insecurity.pdf
see release-critical bugs at http://bugs.debian.org/bugs/release-critical
Where have you seen transparent quality control like that at M$?
AI and quizzing are just small parts of the value of computers in education.
Many school systems are concerned about the cost of computers but with FLOSS and older/lower-powered client machines, schools can run powerful Linux terminal servers for less than half the cost of that other OS on a fat client. With donated PCs wiped of that other OS and booting via PXE/LTSP, a school only needs a good network, powerful servers and a bit of care to set up the system for the different way of operating.
We had twice as many seats installed this way as we could have afforded with that other OS and almost nothing to do for maintenance except backup and watching it hum.
What those who compare LTSP and such with dumb terminals of 20 years ago is that Moore's Law had dragged the CPU, RAM, networking and motherboards into the 21st century. A modern CPU is idling on most desktops (80%) and RAM is cheap and plentiful on 64bit servers. LTSP is a compromise only for full screen video and heavy number crunching. Even then, a power user gets a slice of monstrous power on a multi-CPU Optern server loaded with RAM instead of a typical desktop. The school's load is almost ideal for LTSP except if you teach multimedia production in a lab. Those 20% of users who need it can still have a Linux thick client but that solution is unnecessarily costly for every seat.
A similar technology that multiplies benefit/cost is the multi-seat X client. You can hand 6 to 12 users on one PC using single or dual head video cards and multiple keyboards and mice. Groovix and Userful are two implementations. Groovix is open source.
There is reason to believe that there are 1000 million PCs in the world and 200 million new ones are being produced each year. Even if folks were willing to switch to Vista, it would take five years and nearly a trillion dollars would be spent on this Vista foolishness. That might make sense if there were a real benefit to Vista. Consider the alternative scenario. Instead of junking 200 million PCs each year, migrate them to GNU/Linux and they keep running. This migration will either quickly double the number of PCs running in the world or slow down the migration to Vista as being unnecessary. This is not like going from a Pentium I to a Pentium III or some such upgrade. Current PCs are mostly Pentium III and better, working very well. It makes more sense to migrate working PCs to Linux than to chuck them and go to Vista with unnecessary hardware.
People know their PCs are good enough because of what they can do with XP. They will not be fooled. When XP is no longer supported in 18 months, the world will be overrun with malware as the XP machines stay on-line. Folks will have to migrate to GNU/Linux.
M$ itself admitted that Pentium III or so was good enough for clicking and gawking. See http://www.usdoj.gov/atr/cases/exhibits/365.pdf
The world and Microsoft have spent billions getting the stuff to work. Then Microsoft kills it and the world has to buy new equipment, pay Microsoft more money and start all over again. There would be no objection to Microsoft continuing to improve a product that many find useful and there would be no objection if hardware could be continued in use for its working life instead of the life of Microsoft's product. The reason that Microsoft cannot continually improve its product is that it is spaghetti code (not modular) and everything changed has unfortunate unintended consequences affecting other parts. That particularly ticks off developers who know how code should be developed. On top of the life cycle issues, the poor design and large changes to the new product introduce a new set of vulnerabilities that drive admins batty.
Look at the recent release of Debian 4.0 Etch. It could be a large jump from Sarge, but it is not because Sarge users could follow the developments in testing, watch bug reports in Etch and make the change when it suits them. OSS is just a better process. Watching hundreds of millions of PC users go down the wrong path is like watching genocide. Good people just have to say and do something about it.
Last year I installed a new system based on LTSP (http://ltsp.org/) on EdUbuntu. It was a few days before school started and I had the staff in my computer lab to show them the basics of the new system (teachers are usually not adventuresome like kids...). I was a few clicks into the presentation when a display driver started corrupting data eating away at the bottom of the screen. It turned out that good old Vesa driver did the job, but I did not figure that out during the presentation. What a flop I was! My boss dozed off! When school opened everything went smoothly except I did not have LDAP properly configured so some users had different passwords on different servers (not Debian's fault) and some memory going bad in one server. There are times when the software just has to work and that is much more likely with something that has existed in testing for quite some time.
It is too good to be true, but this is not an April Fools joke.
etch has been less buggy than sarge for almost a month. :) ... and less buggy than that other OS since Etch began to boot. It is absolutely amazing that a loose, democratic organization like Debian can out-perform a giant of industry with giga-bucks in the bank and tens of thousands of slaves but it is true, the bazaar works. Freedom and empowerment take some work and spine but the result is worth everything many times over.
Debian is one of the great old distros that just keeps getting better and not by adding frills. It is a large distro on many architectures supported by package managers from around the world. It is not hard to install as the reputation was. It is huge with many thousands of packages all smoothly (well, mostly ;-) integrated. I favour it for anyone migrating from that other OS, a new installation or on a large or small system.
One of the neat features of Debian Etch is the smooth set of packages for installing LTSP (See http://ltsp.org/ ). One can go into a school on the weekend, set up a server and support all the old equipment as thin clients whether they be iMacs, i386, i486, P-what-evers and manage hundreds of accounts by Monday.
I have been using Testing for a couple of months and there are few bugs. Nothing has prevented me from using it in production.
Congratulations, Debian.org!
Customers are demanding service from hardware manufacturers for Vista drivers, not from Microsoft. HP and others want to sell new hardware not patch old drivers. The customer wants his working device to continue working.
It is too late now. They have burnt too many bridges. Even the hardware makers hate MSFT because they changed VISTA just enough to break all the drivers. Millions of school kids are experiencing the richness of GNU/Linux. Dell and HP are getting serious about Linux machines (maybe). The EU may realize that mega-fines are not enough and outright ban MSFT. Even Uncle Sam is discouraging the use of Windows for security. MSFT is like a long-necked dinosaur trapped in a tar pit. The brain is so small and so far away that it has not received the final BSOD yet. The body is cooling slowly because of the large mass to surface ratio.
Not that I think MSFT will learn to be more competitive, and they have burned too many customers, offended too many governments, produced too much schlock, told too many lies and cost the world economy too much to be welcome in the market place.
I strongly advocate that every geek drop in at a local school or school division and talk to them about IT. The existing clunkers may make fine thin clients and a few new servers do not cost much to use as Linux terminal servers (see http://ltsp.org/ ). There are several distros that automate the conversion of a reasonable PC with two NICs, some extra RAM and storage into a Linux terminal server so that the old machines or new thin clients have only to show the pictures and receive the clicks. Thus a whole lab gets to log in and run on a single good server. Maintenance is down an order of magnitude this way as only one machine needs a file system. Debian, Ubuntu, SkoleLinux and K12LTSP have reasonable repositories for schools. These system do everything a school may need except allow full screen video to every seat. All the normal click/gawk/click stuff works beautifully. Teachers can easily monitor/control each student using VNC or whatever. I like to log a student out when they wander... or I block their favourite time-wasting sites. These are powerfull payoffs for schools who make a small investment in effort to install such a system. It is useful in a single classroom, a lab, or a whole building.
Schools that have a bit of money to spend can invest in a powerful server that can run a whole school over gigabit/s. A motherboards like TYAN S3992 (see http://www.tyan.com/product_board_detail.aspx?pid= 235 ) are just made for this with Lots of RAM, dual gigabit/s NICs, and dual Socket F Opterons. The investment in such a server is spread over every seat in the system and the per-seat cost can be $25 or so for the server, free systems donated by government or business ($0) or new thin clients such as NTAVO 6040 ($139) with LCD screen, USB keyboard and mouse really is very cost effective and easy to maintain. Having a single server is a little less reliable as a single point of failure but having fewer parts also saves money.
I have found Windows to be exceptionally secure after I install GNU/Linux right over top of it. I have never had a failure after this procedure. I started doing it when I saw machines running Windows fail for no apparent reason, sometimes just idling and "PFFT!", dead as a BSOD. People told me it was hardware problems, but, running on the same hardware, the new installation would run for months with no downtime.
Sounds like "Inactive Directory" at work. On my last system, after a few requests from folks who needed to change their passwords but did not know how, I typed a single line command that put an "apple" icon on every desktop with the words "change my password" underneath. That change went out to four servers on encrypted connections and it was so, just like God at Creation. Those who came to me in person had their passwords reset and they sat down at a chair next to my terminal and logged into their own configuration and changed their password to suit them. With folks in a WAN, that would not be possible and there would have to be some levels of filtering to prevent hackers from phoning up and asking for Joe's password to be reset... I created 700 accounts in a few minutes on my system with random passwords and usernames and handed books of coupons to managers who issued them to underlings and recorded the transactions. The whole user system was set up in an hour at my end. I detected two of those accounts being misused and suspended privileges instantly. No sweat.
Linux is the right way to go. With simple scripting one admin can control/monitor hundreds of servers with similar configurations. Of course, he can futz it up pretty quickly, too, but that is why Linux sysadmins draw top dollar.
I switched to Linux when I had just five Windows machines that crashed daily just when someone needed them to perform. I put Linux on out of desperation and the same hardware ran six months without downtime. People who use that other OS just cannot get their minds around a system that works, has few bugs, costs little, and is flexible enough to do whatever we wish.
Many cannot imagine watching a programme run on one machine from another machine. In the Windows world, that is two points of failure. In Linux, it is the way the display system works. You can have as many displays as you have resources and they do not have to be on the same machine. Bandwidth is only an issue if everyone is watching movies instead of working...
The truth is the typical thick client is idling at a small percentage of CPU load waiting for a mere human to read and click. The idea of server centric computing is to give such unused power to the next guy. My AMD64 3000 (1.8 gHz clock) can please a whole room full of people (I have run 30 simultaneous users) on this basis alone. If you add to this caching of files, huge buffers, and RAID storage, you should ask, "Why would anyone want a pokey thick client that needs software and hardware maintenance and replacement every few years?" My server cost $1200 to build (about $50/client). Programme loading and file serving uses no bandwidth. I have everything on the server. I have gigabit/s bandwidth and 30 megabits/s is plenty for one client.
For 3500 clients, you could install 120 servers like mine of 30 quad core machines or 20 quad machines with faster clocks. The multi-opterons do cost more but you can put way more than 4gB RAM on them.
In case anyone is wondering, I run Linux, not stuff designed to run a single user.
The last system I designed had 130 seats as Linux thin clients and I could tweak the whole system without leaving my chair in seconds. I had redundant servers ($1500 each) instead of redundant clients and it took only minutes a day to verify that everything was OK and it was for months. Not one incident of malware disrupting anything. The users migrated from needing a full time geek to re-install that other OS several times each year on each client to having machines as reliable as telephones.
Give me LTSP or even a remote X session any day. My files stay on the server. Only pictures of their contents and my desktop follow me around. Makes a lot more sense.
Moore's law increases the power you can put in boxes in the server room so rapidly that you do not even have to increase the size of the server room. Just keep upgrading the servers. A few years ago we jumped to AMD64. Computing MFLOPS did not increase much but throughput sure did. Now we have higher clock speeds and multicore chips. Server-centric computing just makes sense. There is no need to upgrade the thin clients for many years and yet we can constantly redouble our computing power in the server room. So what if we have to water cool them? I like to keep all that noise and heat away from my desk.
I disagree strongly with this statement:"The cost savings of opting for a truly non-capable display-only device over a competent computer is essentially non-existent."
With Windows it is true, because Bill charges per-seat. With Linux you get about 66% discount per seat. The cheap thin client box may cost only $150 USD. You pay Bill more than that in licences.
I recently equipped a whole school with thin clients in every room and we got twices as many seats as we could have afforded with that other OS and the cost of maintaining the thin client is almost zero. Over ten years, they will save their own cost in power savings, so they cost almost nothing. On the other hand, a thick client costs over and over again and is a burden rather than a blessing. There are a few things a thick client can do that a thin client cannot, but that can be worked around: go thick for the few individuals who need thick. Why pay big bucks to have thick clients idling all over the building? Pay big bucks for servers working hard.
If businesses A and B have 100 seat computer systems and A pays tens of thousands of dollars for software and B uses FLOSS, B has an advantage. For seats that involve the usual office stuff (word processing, spreadsheet, image processing), GNU/Linux works today on the desktop.
I think most of the "special" apps with the high price tags are archaic and represent lock-in. What do they do that requires a particular piece of software and no other? I have heard many say that they can work with PhotoShop but not with Gimp and I ask for details and get some tiny detail that is an insignificant part of the task.
I was once involved in a project that was locked-in with software that took six man-years to develop. The developer went bust. I replaced the software with all the functionality we needed in six man-weeks of writing from documentation and the result was better. We used to train employees to use the old programme (and the industrial systems it controlled) in six months. The new programme took only one month of training because the user interface was greatly improved (hints at the bottom of each screen). The entire cost of the transition was recovered in the first year because it took so little time to train users. This was mission-critical software without which the system could not run. The cost of the whole system was millions of dollars and it was in jeopardy if we needed to change anything in the software, so we replaced it with something we controlled. It made business sense, was cheaper, and it worked better.
I recently moved a school from Windows to Linux with no downside. Everything works. We control our destiny. We saved a ton of money on each seat and could afford twice as many seats as was planned when Windows was the concept. Linux is ready for the desktop certainly for many users who browse, write, move and store stuff.