“in its home country of Germany, according to the latest independent study from ICT market analysts, ama GmbH. In the survey of over 6,000 businesses with more than 50 people, thin clients accounted for 11% of desktops and 35% of them were IGEL devices.”
see Survey Reveals IGEL as Clear German Market Thin Client Leader.
10% of desktop PCs being thin clients seems small but it is not. They last three times as long as thick PCs and they can run GNU/Linux instead of that other OS. That’s huge, a potential 30% loss of share for Wintel. That’s right; thin clients don’t need to be x86. They can be ARMed as well.
While M$ does attempt to tax thin clients with huge server licensing fees and CALs, more thin clients run a browser and access web servers directly, cutting M$ out. The annual rate of production of thin clients is still quite low but as Wintel PCs decline in number, the stark reality of the platform declining in all directions looms on the horizon. M$ has nowhere to go but down on client OS and their server licences depend heavily on that installed fleet which is about to shrink. All is not lost but there’s no upside for M$ for the next few years. Small cheap computers of all kinds are eating M$’s lunch.
17408
12734
206
3
2
23843 
11832 
11684 
4615 
4236 
1641 
198 
14 
2 
0 
0 
0 
Robert, I think this signifies where desktop computing is headed towards. With a web browser the only requirement, applications are server based. The end user’s interface only need to be a net ready appliance. Then a company’s set of servers essentially become the local “cloud”.
Desktop blade or appliance malfunctions? No problem, IT staff swaps it out, reconnects monitor, keyboard mouse, powers on. With minimal setup, the user’s service is restored.
Without an appliance hard disk required, the “FAT” is trimmed (pun intended).
George Hostler wrote, “Desktop blade or appliance malfunctions? No problem, IT staff swaps it out, reconnects monitor, keyboard mouse, powers on.”
I think in many organizations, the local secretary/phone-person will have a unit or two on a shelf with detailed instructions on how to plug things in. Folks think thick clients die at a rate of a few percent per annum. Thin clients can go an order of magnitude lower rate. I installed 96 units in one school and after six years not one has failed despite proximity to kids.
Robert Pogson wrote, “Folks think thick clients die at a rate of a few percent per annum. Thin clients can go an order of magnitude lower rate. I installed 96 units in one school and after six years not one has failed despite proximity to kids.”
Overall with teachers supervising, the kids tend to behave themselves and the wear and tear on these units IMO are less than in a business environment (for example, someone who does journalism or technical writing, or much correspondence).
Regardless, the nice thing about these blade units is usually they have no moving parts. Without a hard disk to go bad, about the only thing to go on a high usage unit would be the keyboard, especially if the keyboard was inexpensive light duty one. However with those items under $20 US are an easy replacement.
Thin clients do have there limits without insanely fast network.
George Hostler
–Overall with teachers supervising, the kids tend to behave themselves and the wear and tear on these units IMO are less than in a business environment –
Depends on the business environment and the school. Thin clients and thick network clients have a habit of out lasting the tradition fat client with local hard drive in schools. Comes down to two things one shock resistance. Yes I know hard-drives are meant to tolerate so much G impact before screwing up. Chair hitting computer because kids are fighting can at times generate more force than drives are design to tolerate. Teachers are not always 100 percent in control of students. Fights near computers are less in the business environment(not impossible).
The number 2 is the thin/thick client can have no where a lazy kid can insert garbage into machine instead of walking to bin.
–Regardless, the nice thing about these blade units is usually they have no moving parts.– Basically that feature makes the item very good for students and business.
Moving parts add another failure point. Or in the case of sawmills somewhere that will fail in a week if its in the cutting room if its not like a hard-drive well filtered but that will only give you a few years. Thin clients full sealed in that environment last really well.
The more hostile the environment the less suitable a tradition PC is. Now a solid state arm tablet that does not need fans and venting also is highly suitable in hostile environments.
More heat something electronic circuit generates the more cooling it needs the more likely it will fail. We do need a race to low heat generating.
oiaohm wrote, “Thin clients do have there limits without insanely fast network.”
If you do not do audio/video over the LAN, thin clients are actually faster than thick clients because one can afford to install one fast file-system on the server and huge RAM. I’ve measured the bandwidth used by a thick client with file-serving and it’s more than with a thick client for text/images because most of the data is kept off the LAN, also a security-feature.
pogson it gets hard. I do mean it gets hard comparing thick and thin.
Thick is more tolerant in a congested network. Since the Thick can do stuff without needing to refer to server.
Multicasting with Thick can be used to reduce bandwidth the server needs to provide. So a single thin client vs a single thick client the thin will in most cases win on bandwidth. When you start looking at a 100+ in a network. This may not be true the thick might be using less network bandwidth from the servers/network storage. Of course the thick will require a decent block of ram to cache the network provided files so it does not have to request them. So a thick client requires more ram than a fat client doing the same job. This use to be a problem when ram was expensive and not large enough to store most applications fully in ram.
Please note I use thick thin and fat. Fat being local hard-drive. Thick and thin are both network providing. Thin applications server side. Thick applications client side.
Thick you are still after a faster file-system on the server right? Maybe not. Thick is one of these more interesting odd balls with Linux of course.
Cluster file-system can host the root image of Thick. So you have a stack of fat clients that are the file server for a stack more thick clients. Only 1 machine in the network need to process updates since the next time the machine reboot the updates will be applied.
A fat/thick network gets very interesting. Because there might be no such thing as a central server. Yes a fat/thick does not conform to any model Microsoft does. fat/thick done right is highly fault tolerant.
It is possible todo thin/thick like the ltsp does. Its possible todo fat/thick like disk less remote boot does. So far I have not seen anyone pull off thin/thick/fat. Basically a server-less solution doing all 3. Might happen one day.
Its the central server point is where you need huge in thin. Thick with server the server has to be decent. Thick with cluster file-system with fat clients providing as long as there is enough fat clients to provide the cluster file system fast enough everything is fine.
Yes the thin security feature of data kept of lan is a reason todo it at times.
pogson basically its a complex problem.
Really with all the possible different solutions to getting interface in front of client the market will break up.
Some sections will be care for well by thin clients. Very well.
Some sections will be care for well by the mix of thin/thick clients.
Some sections will be care for well by mix of thick/thin/fat in a server based solution.
Some will be served well by thick/fat networks.
Then some will be served well by the likes of chromeOS and Android and other portable devices with a server.
Problem is this kind of market breakdown does not suite Microsoft.
Not only are Linux thin clients lightweight, economical, and robust, but it is easy to add bandwidth and capacity to the Linux server.
Linux readily allows cluster computing, all you have to is add boxen and they can be tightly integrated to act as one giant multicore machine.
Linux based RF routers can be set to multiple channels. Need more RF bandwidth, just add more low cost routers. After a dozen or so it becomes less easy, but that is plenty of bandwidth for almost all schools and office buildings.
ram I still prefer wired where I can prevents the nightmare of a noise welder or other big nasty noise source ruining your day.
Yes I do agree with you. Big thing is all the different configurations I mentioned Linux can do without being a major license annoyance or other problems.
–Linux readily allows cluster computing, all you have to is add boxen and they can be tightly integrated to act as one giant multicore machine.–
Linux support many forms of cluster. Individual processing node clusters as well.
Linux is flexible to suite the space. The Linux question is more what shape hole is the problem and what is the best peg I have. Where with windows is one we have the square peg how can we make it fit this odd shaped hole. Problem is almost no problem hole out there is square.
Basically I am not saying Linux fits perfectly. The issue is with Linux you have more options to attempt to get a perfect fit.
Don’t forget zero clients. Thin clients aren’t the only ones in the game.