virtualbox.org

when you look to the most common office use of computers they are used for lightweight stuff like editing documents. so why we don't set up one relatively powerful workstation with Dropbox and connect 10 or 20 monitors , keyboard and mice (via ethernet) for employees allover the office to use?

does this technology exist? if no why not ?

I remember back in Windows 95 days someone sold a card to attach multiple keyboards, mice, and monitors to one PC. If it is still sold, it has issues. A google on "multiple monitors keyboard mouse one computer" (it's like the old phone company commercials from the 80's, "Phone First!", now it's "Google First!") shows this post: http://www.tomshardware.com/forum/44245 ... e-computer, with possible problems with such a setup, as opposed to individual PC's.

Virtualbox-wise, if you can get USB keyboards and mice with different manufacturer designations so the USB filters could be set to match each keyboard and mouse to a guest, then you might be able to do such a setup, on a three- or four-guest scale. 10-20 guests will require a super-beefy multi-thousands host, with one point of failure, as noted in the link above.

If you're set on one host PC, use "thin clients" - small PC with keyboard, mouse, monitor, something like ThinStation for an OS, and Remote Desktop to remote into the guests via network.

You'll probably spend more getting a 10-guest server, individually-licensed guests, and thin clients set up than if you bought individual refurbished warrantied pre-OS'd PCs.

People keep trying to reinvent the minicomputer!

The question IMHO contains a logical fallacy. If the problem is lightweight then any halfway capable terminal would have enough processing power to run the software directly, and without the single point of failure and budget hogging of the timeshare solution: which is how we got to personal computers in the first place, from the timeshared minis or mainframes which came before.

There is also the question of software licensing, but basically any license that allows one mode probably allows the other.

mpack wrote:People keep trying to reinvent the minicomputer!

The question IMHO contains a logical fallacy. If the problem is lightweight then any halfway capable terminal would have enough processing power to run the software directly, and without the single point of failure and budget hogging of the timeshare solution: which is how we got to personal computers in the first place, from the timeshared minis or mainframes which came before.

There is also the question of software licensing, but basically any license that allows one mode probably allows the other.

lightweight is not the problem ,the problem is that those $500 computers can do more than what many offices need. so why don't take advantage of that?

after i wrote the topic yesterday an idea came to my mind : raspberry pi . the raspberry pi 3 is powerful enough i think. if it is not in the near future i think 64bit ARMs will have that role.

So why exactly is VirtualBox a better solution compared to a Terminal server? Why would you want to add another layer of complex setup? Just out of curiosity...

docesam wrote:does this technology exist? if no why not ?

It doesn't exist for a couple of reasons:
• It's convoluted.
• It doesn't improve upon existing solutions.

Don't try to find a problem to apply a solution, try it the other way around

BTW, VirtualBox and ARM; no gonna happen.

If the raspberry pi is strong enough to handle remote desktop, it may serve as an inexpensive thin client. You'd still be left with the supercomputer to run the environments. Terminal server may be a good idea.

Though, the pi's don't have much in 3D graphics, I'd guess, so one day when the office starts to need that fancy program because the business has done a paradigm shift, and stronger PCs are needed on each desk, then the supercomputer may be overkill.

scottgus1 wrote:Though, the pi's don't have much in 3D graphics, I'd guess

In fact they have a whole graphics core running alongside the ARM - though "core" should be taken with a pinch of salt, since I believe it's some kind of vector processor. There was also a difficulty with libraries when I last looked. Broadcom make a lot of money selling those SOC chips to Apple, they don't particularly want to give the good stuff away at RPi prices...

socratis wrote:
BTW, VirtualBox and ARM; no gonna happen.

i didn't mean that we run virtual box in ARM ,i mean that the pi can replace a whole PC as very lightweight document editor / internet surfing machine. with the evolution happening in ARM i anticipate this will happen sooner that you think.

I still fail to see where VirtualBox fits into the grand scheme of things compared to more mature, already deployed solutions. As I said, "Don't try to find a problem to apply a solution, try it the other way around".

socratis wrote:I still fail to see where VirtualBox fits into the grand scheme of things compared to more mature, already deployed solutions. As I said, "Don't try to find a problem to apply a solution, try it the other way around".

50 employee office , $500 computer = $25,000
50 employee office , $200 raspberry pi = $10,000

that is only 50 emplyee office. do you think this is a significant saving?

So, what is it exactly that you're asking from the VirtualBox users and developers. That's the part I don't understand. To plan/design this thing?

this is a discussion , at first i thought since there is no lightweight cheap device virtualization could save money. but as we went along the discussion i realized maybe the lightweight cheap device is available or soon to be available. so (at least my ) conclusion that VB is not the optimal solution for this.

At the moment Virtualbox does not have a scalable method to take multiple keyboards, monitors and mice connected to the host and distribute the devices to specific guests. 2 or 3 from different manufacturers, maybe. 50, highly unlikely.

Virtualbox can distribute guests via Virtualbox Remote Desktop Server, or the guest OS's can distribute their own remote desktop streams. A $35 Raspberry Pi can apparently log onto those remote desktop streams (http://liliputing.com/2012/05/raspberry ... sktop.html). You just need a Virtualbox host capable of running 50 guests simultaneously. 128GB ram at least, a dozen raided SSD's, dual or quad multi-core processors, etc. Big "cake"* for that host, but the guests are easy, assuming you can license the software.

*As Carl calls money on "As Told By Ginger"

docesam wrote: 50 employee office , $500 computer = $25,000
50 employee office , $200 raspberry pi = $10,000

Yet again I refer you to the minicomputer analogy, because I find your budget estimate hopelessly naive. If you have a central computer system then you have an office and support staff dedicated to looking after it, and the salary of the tealady who looks after that office will more than swallow up the savings you envisage.

And what about lost revenue when your timeshared system is down for maintenance? That 50x multiplier applies there too.

Believe me, if timeshared computing was the answer then personal computers wouldn't exist. It only made sense when buying and operating a computer was ludicrously expensive. As in, if you aren't a government or a megacorp then you can stop looking.