The desktop monopoly of M$ is looking very old today. Emerging markets are consuming an ever larger piece of the PC pie (Asia/Pacific exluding Japan shipped about 1/3 of PCs with per annum growth of 11%) and they are not loyal servants of M$. Consumerization of IT means stuff consumers drag in to work will have to be accommodated. No more “M$ shops”.
The numbers are spectacular. Last year folks were expecting the world to ship 45 million tablets and revised estimates to 60 million. It turns out Apple, on its own, shipped that many. The global shipments of tablets were close to 100 million. And then there are the smartphones which shipped in greater numbers than desktop/notebook PCs.
All this means M$’s share of personal computing is plunging like a stone and more importantly, the share of people who look to M$ for the source of software is declining rapidly. Only a tiny percentage of smart phones and tablets use M$’s stuff. Soon the desktops and notebooks will not be using M$’s stuff. The change in 2012 could be dramatic as more businesses use thin clients and/or web applications and have less dependence on M$’s software. In the last quarter M$’s client division decreased its revenue 6% on the basis of December’s shopping mostly. There’s no sign of a decrease in consumer spending on non-M$ software. Shortages of hard drives will impact desktop and notebook PCs seriously in 2012 on top of everything else.
Still, there are many who are seriously locked in but they may have taken their last step on the Wintel treadmill.
9628
8997
102
2
0
13236 
5986 
5948 
3836 
1690 
1553 
199 
1 
0 
0 
0 
0 
The Apple monopoly is quite as bad as the Microsoft monopoly, and possibly worse. In fact, Apple’s walled garden/company store approach in combination with the cultlike “elite” loyalty the company inspires may well prove an unbeatable economic combination.
“Only a tiny percentage of smart phones and tablets use M$’s stuff.”
Currently correct, yes.
“Soon the desktops and notebooks will not be using M$’s stuff.”
And that would happen exactly why ?
“as more businesses use thin clients”
Evidence for that happening ?
“M$’s client division decreased”
and their business division increased, which doesn’t seem to tally with your point about thin clients.
Thin client usage is rising rapidly.
“record year in 2011 where the business saw strong organic growth in sales of its Universal Desktop thin clients and 570% growth in license sales for its Universal Desktop Conversion software, designed to convert old PCs and thin clients into easy to manage IGEL desktops.”
see IGEL Partner Days: Staying Ahead in the Thin Client Business
M$ takes away much of the advantage of thin clients by replacing an OS licence (say, GNU/Linux) with a CAL required by the messy EULA which limits connectivity arbitrarily. Large roll-outs of thin clients by businesses and other organizations make less sense with that other OS. With GNU/Linux, one avoids the licensing fees and CALs, making the cost of a roll-out much less. When I did such a roll-out, my savings on licensing was the deciding factor for my employer, but the reduction in labour required was a huge blessing. At the same time, performance was improved because it’s the performance of the server that’s the limiting factor.
“IGEL”
Completely irrelevant.
“my savings on licensing was the deciding factor for my employer”
For businesses, the license costs for MS are negligible.
“reduction in labour required was a huge blessing”
Right – if you can get away with the restrictions you thus impose on the clients. I see how that would not only be practical but a good idea in a school, however you won’t always be able to get away with that in a business. (In the company where I work, we use some terminal servers for remote dial-in, but it’s fat client for everything else.)
“performance was improved because it’s the performance of the server that’s the limiting factor”
A modest desktop PC can be had for little more than a thin client and is easily able to run stuff without a server backing him up. A server with, say, 10 times the computing power of a desktop PC is likely to cost money – especially if you get a real server because after all the server in a thin-client scenario would be a single point of failure.
Hello Mr. Pogson, what do you think about this report made by Google who says smartphones and tablets aren’t replacing PCs?
http://www.theverge.com/2012/1/25/2732103/google-smartphone-usage-statistics
Google may have some smart people but they are mathematically challenged. If PC use is holding steady while smartphone use is rising, then PC use as a share of computing is declining…
We have seen for several quarters now double digit growth in smart thingies while x86 PCs are flat. I believe the use of x86 PCs has peaked. No one needs a more powerful computer than a smart thingy for personal use. The few that do render video etc. do not make the market in PCs. The few that use QuickBooks or PhotoShop do not make the market. However, everyone and their dogs will soon have smartphones. I know several people who use mostly a smart phone. They may never buy another “PC”.
Ch wrote, “For businesses, the license costs for MS are negligible.”
Depends on the business. Call centres for instance want machines that are trouble-free, cheap and plentiful. They don’t need a desktop PC nor a licence for that other OS to do the job. A thin client works and it can cost less than $100 running GNU/Linux. Do you actually believe a call-centre would be set up and plunk for M$ by default?
We have seen a lot of migrations and many of them were businesses ranging from Ernie Ball Guitar Strings, to lawyers’ offices, to Peugeot. Cost is a factor, a big one although security and ease of use may be others.
Ch wrote, “A modest desktop PC can be had for little more than a thin client and is easily able to run stuff without a server backing him up. A server with, say, 10 times the computing power of a desktop PC is likely to cost money – especially if you get a real server because after all the server in a thin-client scenario would be a single point of failure.”
Completely irrelevant. A typical desktop has one hard drive and so seeks slower than a server with 6-8 hard drives. The server can seek several files simultaneously and likely has far more RAM making file caching much more effective. I build GNU/Linux terminal servers that blow away desktop PCs for about $25 of hardware per user. That takes account of the fact that typical desktops have CPU utilization less than 10%, a terrible waste. Price/performance for thin clients on a terminal server are much better than desktop PCs.
A big roll-out of thin clients well documented may be Largo FL, where they deploy huge servers costing ~$40K but serving 400 clients simultaneously, only $100 per client. Combining that with a $100-$200 thin client box gives a huge increase in performance/dollar with much less maintenance.
One big hard drive, PSU, motherboard and one big CPU per user is a ridiculous waste.
@Robert:
“One big hard drive, PSU, motherboard and one big CPU per user is a ridiculous waste.”
Is it? Why?
Let’s assume you can build and distribute a thin client alternative for free. The fat client alternative will cost you maybe $250 per year (amortized over the standard four, and including business software that a call center operator probably wouldn’t need). I’m probably erring on the expensive side here.
So, that’s $250 you didn’t need to spend.
But you’re hiring the staffer at $15,000, minimum. You’re paying for office space. You’re paying for toiletries and heating and local taxes and God knows what.
And after all that, you want to make your employee feel as wanted and as important as you possibly can.
You tell me, Robert: to save a (purely theoretical) 1.25% of your labour cost, why would you want to piss your staff off by sending them the signal that they’re back in the 1970s sitting at a 3270 terminal?
Have you tried to sell this concept to any local businesses? Because if I am wrong and you are right, you could make a tidy packet on consultancy here.
Dr Loser, not understanding efficiency at all wrote:
“One big hard drive, PSU, motherboard and one big CPU per user is a ridiculous waste.”
Is it? Why?
Because you can get the same performance or better without them. Let’s calculate the capital cost per seat:
Total is $255. For the usual thick client that would be at least $200 more, I would bet, and you still probably want a server. A server with that other OS costs $thousands to start.
Then there is the cost of operation. Maintenance of one server compared to ~100 thick clients is a huge savings.
Then there is performance. The server might have faster/more drives, more RAM, faster NIC, and it puts out heat and noise somewhere else. The RAM is not to satisfy each user. It takes only 40-100MB to please simple users. The huge RAM caches stuff so the files needed to open a browser window or to start the word-processing come from RAM, many times faster than rotating magnetic storage. Speed-up I have seen: 3X for applications, 5X for logins. Users appreciate that.
So, getting more for less is a valid business objective/principle. It increases the capital available to invest in productive stuff like salaries. The amounts are not trivial. Multiply $200 by 100 clients and you’re talking about $20000, a few days’ pay for the staff or a bunch of great printers.
“Users appreciate that.”
But users do not appreciate going backwards in functionally. I guarantee you Pog that you would have discovered this first hand if you ran into an environment where people did more than the simple tasks you are used to.
come to think of it…
Perhaps you did already in one of those places that you “moved on” from when they didn’t just let you pave over the existing environment with linux, eh?
oldman wrote, denying reality, “But users do not appreciate going backwards in functionally.”
Some users will go backwards switching to GNU/Linux on thin clients but many will go forward. What about them? Many millions of thin clients are bought annually and not by uninformed consumers.
oldman wrote, “Perhaps you did already in one of those places that you “moved on” from when they didn’t just let you pave over the existing environment with linux, eh?”
I have once or twice moved on because of the way IT was handled in schools but that was rare. Many employers were quite appreciative that I could keep things running smoothly without using M$’s junk.
@Robert:
Me: Is it? Why?
You: Because you can get the same performance or better without them.
You are utterly impervious to argument, aren’t you, Robert?
I don’t mean you never accept defeat (not relevant in this case). I don’t even mean that you pull an oiaohm and roll off on tangents. You don’t do that, in general. You just repeat whatever you said before.
What I mean by “impervious to argument” is that any other person’s opinion, however ill- or well-founded, is like water off a duck’s back to you, isn’t it? You literally fail to acknowledge its existence.
My entire argument was about the relatively minor savings involved and the huge insult that a large company would be offering its employees if it chose to go down the thin client route. I was even kind enough not to mention the hidden costs, the traumatic cross-over period, the mess, or even to question your bizarre assumption that things would be “as good, or even better.”
Really? Over a not-necessarily always-on Internet connection? Using, for the most part, Web clients?
But, OK, I gave you all that. These things can be dealt with on a rational basis, further down the road.
For now, all I was saying is that
(a) A company would have to be far more evil than Microsoft to treat their employees this way for the sake of a few dollars here and there
(b) Almost nobody does it, which must mean (if it is a good idea) that you could make a tidy living as a consultant. After all, you know this stuff inside out.
Did you bother to address any of this? No.
I don’t know who was responsible for letting you into the wonderful world of education in the first place, but at this point it seems that whoever did so was a pretty rotten judge of character: whatever other qualities you possess, being able to listen to other people is clearly not one of them.
You’d make a pretty good cabbie, though.
“What about them? Many millions of thin clients are bought annually and not by uninformed consumers.”
Consumers don’t buy thin clients Pog. Some businesses and institutions do. But if you were to really look, most of the thin client sales are to go into systems running the same desktop software that they were running on windows. IN those cases thet users are not going backwards., If anything they are being offered the same software in a way that can be managed centrally. Of course I realize that you will never accept delivery on the notion that FOSS on linux is going nowhere on the majority of thin client implementations. But that is your problem, not mine.
Dr Loser lost when he wrote, “My entire argument was about the relatively minor savings involved and the huge insult that a large company would be offering its employees if it chose to go down the thin client route.”
The relatively minor savings can be huge when you add them all up. That’s why businesses love thin clients. There’s no insult at all to switch to thin clients. Improved performance, better health for the business and better working conditions (noise, heat, space) are all appreciated by users.
“The relatively minor savings can be huge when you add them all up.”
Let me explain the rough dimensions to you:
Cost per employee and month:
Toilet paper: ~ $1
MS licenses: ~ $20
Office space: ~ $100
(highly dependent of location, of course)
Salary: >> $1000
Yes, that are rough numbers, and YMMV, but it should make the relative dimensions clear. So would the employees in your hypothetical company be required to bring their own toilet paper because “when you add them all up” you save a “huge” ammount ?
Ch wrote $20 as the cost of using M$’s stuff per month…
Not even close. There’s probably a few re-re-reboots in there, a few incidents of malware and slow performance. These are huge costs, greater than the licensing. I have been at places that actually cut off Internet access to fight malware. I have been at places where the most anal-retentive software was installed on each PC to fight malware and instead of occasionally encountering malware, users had this monster asking permission to do anything all day long… and of course administrators of systems were burdened with angry users every day at any hour. All that disappeared when we switched to GNU/Linux.
I was a highly-paid professional. Even a few minutes of my time per day to hold M$’s hand in our organization was far more than $20 per month. Same for the users whose productivity was impacted by slow everything. I had one user who used to drink a cup of coffee between clicks because her machine was that slow. I installed GNU/Linux on it eventually and it flew. She had anti-malware software in place.
“There’s probably a few re-re-reboots in there, a few incidents of malware and slow performance.”
Not if you have competent IT staff.
“These are huge costs, greater than the licensing.”
You explicitely wrote about license costs before.
“the most anal-retentive software”
“She had anti-malware software in place.”
And who selected that software ? In the company I’m working at we have a virus scanner installed that doesn’t noticably impact performance.
ch wrote, ““She had anti-malware software in place.”
And who selected that software ? In the company I’m working at we have a virus scanner installed that doesn’t noticably impact performance.”
Ah, some higher-up selected it. The cursed thing calculated a checksum on each application as it started and if it did not match, no joy. Of course updates caused pop-ups all day long all over the building. Of course, I could have wasted hours every week creating new images and propagating them but I had other work to do. This was a highly-rated industrial-strength anti-malware solution and even using it did not prevent malware from taking over dozens of PCs. Malware could still escalate privileges and disable the check for itself and preventing further updates. It’s annoyance was almost worse than the malware. We got the performance hit and the pop-ups.
ch where is the downtime figure. This is the most important figure.
38 week 4 weeks to a Month the cost per hour is under 7 dollar per hour. 1000 dollars a month seams a little low. Since its 152 hours a month for a full time staff that means you must be paying 7 dollar a hour or less.
I would have said at least 1500 dollars ch. So you are at least paying staff 10 dollars a hour. Still this is really crappy pay.
This is why downtime is critical. Lets say you computer takes 1 min to boot. You work 20 days per month with windows updates you get normally about a 4 extra reboots a month. 24 reboots a month. Just in reboot cost usable labour time of people possible waiting is half the OS cost. Why not quarter humans have a habit of being detracted when they have to wait and forget what they were planing on doing. Just in starting up you have cost a lot.
Add fact that background virus scanning slows computer down by a decent percentage. Between 10 to 35(Yes nortons you evil bit works) percent. I will be nice and say the slowdown was only 10 percent. 15 hours in computer productivity per month is lost to the anti-virus. At 10 dollars a hour we are talking a 150 dollar cost. More the staff is paid the worse this figure gets. Since that 15 hours of productive is spread thin users are not noticing the short waits they are doing.
Linux systems can operate from a single read only filesystem holding OS core binaries. That can replicate out very simply. This also prevent root kitting. Since you only need to virus scan the userfiles realtime under linux the scan load drops from about 10 percent to about 2 percent. Yes the read only scanning is like scanning all machines OS cores in one hit. So its 30 dollars vs 150 dollar per month. 120 dollar saving.
Remember this 120 dollar saving double if you run a double shift for the month. Trebles if you are running 24/7. Yes the more hours you are running the computer the worse the number becomes. As your staff numbers go up this grows very quickly. Triple shift 100 active staff per shift. 36 000 a month or 432 000 a year. That is a 300 staff business. At this point is affordable to employ your own developers and develop own internal software to save on some of the overheads.
If you pay your staff 20 dollars a hour you only need 150 staff to hit enough for a programming team.
I have not include a case where the windows machine has got infected yet.
This does not include items like Internet explorer 8 updates not applying correctly so leaving computer non functional either. Yes install Internet next explorer badly and windows explorer the shell of windows cannot start. This is a long term highly annoying defect.
Linux one global audit image option. So you apply update to that image you test that image and every machine gets the same image. Removing update nightmare of random machines failing at random times.
A proper TCO look ch done properly Linux is ahead unless the user is requiring particular applications that don’t run under Linux or don’t have a good enough equal. Even if windows was free it is still behind. There are issues with the ways you must run windows that block it from running effectively on a read only filesystem for the OS core. Linux has no such problem. Minor tech differences make major differences in operating costs.
Ideal is person presses button on computer and under 1 second it comes up and works. This way person does not forget what they are doing. So production goes up. Of course there are very few items that can achieve this.