[besson3c]

Virtualization is in vogue right now. Businesses are finding that they can save literally thousands upon thousands of dollars going in this direction, as are education IT infrastructures. My thinking is that it is only a matter of time before this becomes a way of life for client computing as well.

Once virtualization reaches its potential, the whole idea of having to buy a $1000-2000 workstation every 3-5 years will be replaced by basically buying a screen and a fast internet connection. I know that people have been foreshadowing similar sorts of computing models for a while now, but in the past the hardware has not been up to the job. It is now, and there will definitely be a point where the monthly costs of asyncronous bandwidth over the period of x number of years will be far less than the costs of a dedicated machine. Moreover, this will be more environmentally friendly and far less wasteful. It will also be far easier to buy a computer when you can buy a display for as long as you want to own it, and basically any old dumb terminal from a big box store for well under $100.

My prediction is that Macs will start to look more like the Mac Mini or Macbook Air, minus the need for a fast CPU, expensive cooling, memory, storage, etc.


Here's why I'm thinking this...

Server admins and managers are discovering now that separating hardware from the services they run is far cheaper than having dedicated hardware for each application/department/group. Instead of each department running their own stuff, businesses are building large virtual machine clusters and offering these groups their own virtual machines. The leading VM server stuff comes from products such as VMWare Enterprise Server, Xen, etc. With these sorts of products, RAM can be allocated to each machine on the fly, disk storage can be shared or allocated as needed, and load can be balanced by determining which VM is the least busy and directing new connections towards this machine. VM admins can even do things such as allocate more CPU to certain VMs at certain times of the day, as their need increases - all dynamically and automatically. When the VM environment starts to become sluggish and adding new nodes to the mix does not help, they simply need to add more hardware to the environment.

Contrast this to today's model of computing where you have offices filled with workstations and CPU cycles that are being wasted at all hours of the day, or server side applications that are only needed at certain hours of the day, need to run under a particular OS version that necessitates having dedicated hardware, but the hardware is barely being challenged by the needs of these applications. Think about all of the electricity needed to power each of these workstations in an office that have enough horsepower that a couple years back, the hardware would have been considered top-of-the-line server grade stuff in terms of its performance...

This is definitely catching on in education and big business in building servers, but I honestly think that it is only a matter of time before individual workstations start going this route too. I can see either purchasing a VM along with your internet connection and using a very low cost dumb terminal to connect to your VM, or else having a very simple low cost VM appliance at home for the entire family. Of course, this model is kind of worthless without a fast internet connection, but then again, I feel like I'm at a severe loss without an internet connection now

What do you guys think? Holes in my thinking?

[Chuckit]

If the Internet gets like 500 times faster, maybe.

[besson3c]

I think that when ISPs figure out the money making possibilities here, our bandwidth will increase. There are so many other technologies (such as cell phones/mobile devices) demanding more and more bandwidth that I think it will keep increasing in capacity and decreasing in cost.

[BasketofPuppies]

It will happen, but it's years away.

One of Google's long-term goals is to make software obsolete. Google Docs is the first real threat to Microsoft Office, and it has forced Microsoft, Adobe, Yahoo! and others to get serious about making Web-based versions of their applications.

If Web-based apps replace traditional software--and I believe that this will happen--operating systems, as we know them, will be excessive. Future operating systems could be glorified Web browsers, possibly running over the Internet rather than locally.

Netscape talked about this more than a decade ago. It sounded crazy then, but it was enough to get Microsoft to take the Internet seriously. And now what once sounded crazy is starting to happen.

But, as I mentioned, this is years away. Our Internet connections need to get a lot faster and more reliable before I consider dropping Microsoft Office for Google Docs.

[besson3c]

Originally Posted by BasketofPuppies 

It will happen, but it's years away.

One of Google's long-term goals is to make software obsolete. Google Docs is the first real threat to Microsoft Office, and it has forced Microsoft, Adobe, Yahoo! and others to get serious about making Web-based versions of their applications.

If Web-based apps replace traditional software--and I believe that this will happen--operating systems, as we know them, will be excessive. Future operating systems could be glorified Web browsers, possibly running over the Internet rather than locally.

Netscape talked about this more than a decade ago. It sounded crazy then, but it was enough to get Microsoft to take the Internet seriously. And now what once sounded crazy is starting to happen.

But, as I mentioned, this is years away. Our Internet connections need to get a lot faster and more reliable before I consider dropping Microsoft Office for Google Docs.

I respectfully, strongly disagree.

No backup, no ownership of data is a non-starter for any serious company. Google Docs is not even considered in the enterprise for these reasons. The only way something like this would work is if a company could run the Google docs web application on their servers, or on the Google servers under some sort of legally binding service contract where Google accepts ownership of this data. Either way, Google is not going to do this for free, in which case they are simply competing with Microsoft.

[Visnaut]

The questions isn't if, it's when. Without a doubt it will happen. At first it'll be with businesses, who have a real cost benefit by switching to such systems. Then eventually it will trickle down to consumers.

And honestly, the jump for consumers will be an easy one. With the growing popularity of thin or subnotebooks, and all-in-ones like the iMac, there's already very little "computer" to be spoken for. Even moreso with the line between web apps and local apps beginning to blur.

Owning your own computer, in the sense by which we think of it today, will be like owning your own backup generator. You can get one, but not a lot of people do.

[mduell]

It doesn't work for home users (bandwidth issues for music/photos/movies) and it doesn't work for businesses (data posession, access, etc)... so who is going to use the dumb terminal computing model? Web-only users?

[Sage]

I’m typing this from a dial-up connection – my parents’ house is many miles from any kind of broadband connection. I don’t foresee a web-based OS for this household anytime soon… maybe ever.

[The Godfather]

I hope these promises are true. If more apps are migrated/ported to the "big cloud in the sky", it would mean that my humble 1st gen Macbook won't be obsolete for a few more years.
It is going to get confusing when the 2 proposed models start competing for dominance: 1. the virtual personal computer that could be hella fast when nobody else is using the VM server, or crawling during high traffic, vs 2. web apps a la Google docs, which would enable web 2.0 community features (imagine AutoCAD, FinalCut Pro running in a webserver, delegating grunt work to the CPU farm, getting peer reviews in real time, from any browser).

[besson3c]

Originally Posted by mduell 

It doesn't work for home users (bandwidth issues for music/photos/movies) and it doesn't work for businesses (data posession, access, etc)... so who is going to use the dumb terminal computing model? Web-only users?

It works perfectly for businesses, I'm not following your train of thought... Are you thinking that I'm suggesting that businesses will want to outsource their computing environment to some VM environment? I'm saying that businesses will be running their own VM environment as many are now, and will replacing individual workstations in cubicles with dumb terminals.

As far as the bandwidth issue, it isn't an issue in the way you are describing. All of the processing, storage, *everything* would be occurring under the VM. All the dumb terminal would be doing is displaying the results, simply receiving the signal. We have the technology to do this now in HD TV - maybe the solution here is to give customers a direct pipe to their VM provider?

[besson3c]

Originally Posted by Chuckit 

If the Internet gets like 500 times faster, maybe.

Not 500 times faster. Have you ever done an iChat video conference? The video is pretty smooth, right? It's just a matter of bumping the bandwidth enough so that the video signal can be sent without the compression artifacts.

Again, I'm wondering if a direct pipe to the VM provider would be the best solution here.

[besson3c]

Originally Posted by The Godfather 

I hope these promises are true. If more apps are migrated/ported to the "big cloud in the sky", it would mean that my humble 1st gen Macbook won't be obsolete for a few more years.
It is going to get confusing when the 2 proposed models start competing for dominance: 1. the virtual personal computer that could be hella fast when nobody else is using the VM server, or crawling during high traffic, vs 2. web apps a la Google docs, which would enable web 2.0 community features (imagine AutoCAD, FinalCut Pro running in a webserver, delegating grunt work to the CPU farm, getting peer reviews in real time, from any browser).

The difference is that the VM technology is basically here now. Making a web browser provide the richness of all desktop clients, farming out tasks, dealing with ownership of data in the case of companies, etc. seems much further off.

As far as the issue of loads, it's possible to even it out algorithmically so that planning for the proper infrastructure to support the heaviest of loads is not difficult. As well, it's possible to even dedicate processors to certain VMs, or determine how many processors a VM can utilize. Assuming that storage would be SAN based or something like a SAN, with all of those spindles it is possible that IO performance would be much higher than with home consumer grade hardware.

[mduell]

Ok, I see now that it would be something for businesses to host themselves, not outsourced.

Originally Posted by besson3c 

As far as the bandwidth issue, it isn't an issue in the way you are describing. All of the processing, storage, *everything* would be occurring under the VM. All the dumb terminal would be doing is displaying the results, simply receiving the signal. We have the technology to do this now in HD TV - maybe the solution here is to give customers a direct pipe to their VM provider?

Even on a local network, this is a huge bandwidth issue. 1680x1050 at 32bpp and 60Hz is about 3.2Gbps... even with 10:1 compression (typical high quality MJPEG) and only 3 users per server, you've saturated GbE; go to something more reasonable like 30 users per server and you've saturated 10GbE. If you think the data rate sounds high, consider that I'm trying to maintain a real desktop experience; typical VNC quality is ok for the once-in-a-while session into a remote box, but not for use all day, every day.

[besson3c]

Originally Posted by mduell 

Even on a local network, this is a huge bandwidth issue. 1680x1050 at 32bpp and 60Hz is about 3.2Gbps... even with 10:1 compression (typical high quality MJPEG) and only 3 users per server, you've saturated GbE; go to something more reasonable like 30 users per server and you've saturated 10GbE. If you think the data rate sounds high, consider that I'm trying to maintain a real desktop experience; typical VNC quality is ok for the once-in-a-while session into a remote box, but not for use all day, every day.

Well, there are PCI cards that provide quad gigabit ethernet that can combine their bandwidth (I can't remember the technical name for this), so the problem wouldn't necessarily be bandwidth coming from the server, but receiving bandwidth, and everything along the way.

What do you think about my dedicated pipe idea? Like I said, most people already have this for their HD TV stream they get from their cable/satellite TV providers...

[besson3c]

If somebody wants to start a nice little IT business, they are welcome to steal my idea if they credit me and hire me to work for them

[BasketofPuppies]

Originally Posted by besson3c 

No backup, no ownership of data is a non-starter for any serious company. Google Docs is not even considered in the enterprise for these reasons. The only way something like this would work is if a company could run the Google docs web application on their servers, or on the Google servers under some sort of legally binding service contract where Google accepts ownership of this data. Either way, Google is not going to do this for free, in which case they are simply competing with Microsoft.

I thought that Google offered an option to host Google Docs (and Google Apps) on your own servers. Maybe it was something planned for the future. And while running Google Docs on your own servers might not seem much different than running Microsoft Office over a network, do not discount the ease and convenience of Web browsers.

Regardless, of course Google is going to charge for this. Google is in business to make money, and advertising is inappropriate and distracting in a corporate environment.

[MacosNerd]

Originally Posted by besson3c 

Virtualization is in vogue right now. Businesses are finding that they can save literally thousands upon thousands of dollars going in this direction, as are education IT infrastructures. My thinking is that it is only a matter of time before this becomes a way of life for client computing as well.

Once virtualization reaches its potential, the whole idea of having to buy a $1000-2000 workstation every 3-5 years will be replaced by basically buying a screen and a fast internet connection.

*cough* network computer *cough*

This is not a new idea but virtualization is indeed in vogue. It actually harken's back to the mainframe days and using LPARS. Then it was called the next phase in the computer revolution and the likes of Apple and Oracle were developing a network computer. Most people and business were cool (at best) to the idea.

That being the case I'd be hesitant to proclaim the demise of the desktop (or laptop) computer just yet. I suspect many people like myself will prefer to have their documents on their own computer, rather then trust some ISP or service not to lose or rifle through them. Google does just that their emails which is the number one reason why I don't have a gmail account. .

Businesses may embrace this to a degree but not totally. It makes more sense for them to buy a laptop and allow us to work remotely, in another office with a vendor or customer, etc.

My company is definitely going the virtualization route for its servers because that makes the most sense but not for the desktop and I suspect it may not for the foreseeable future.

[Zeeb]

Virtualization might be great for a company but not for my personal data. I've never liked the idea of paying someone a monthly fee for access to my personal files and I don't like the idea of taking it a step further with virtualization. While this fee might become wrapped up in a monthly broadband bill which I pay anyway I just don't trust it. This virtualization company will essentially "own" copies of my resume, my itunes music, my income tax returns, as well as the software and computing power I need to create new files-- and will be well within their rights to cut me off from access to all of that should I decide to terminate service. Not only would I not have access to my files, but I would not be able to create a new resume with the now worthless dead workstation screen.

[macroy]

From a business standpoint - a lot of this is already going on. You have solutions like Citrix that provide applications to thin-clients. And our organization is already looking at VMware's VDI solution (as well as Citrix's solution) - where an entire virtual client is provided to our users. This gives administrators the ability to provide truely consistent desktops... something that is hard to do in large environments since no one actually replaces ALL desktops at the same time.

Application service providers (what I consider google office types to be) has been around for awhile - the problem with them is that as a corporation becomes larger their need for customization/security and other issue make strict outsourcing very tough. Even in one's own environment, solutions like Citrix have issues with regards to application compatibility. I've not seen one organization running 100% thin applications without issues.

Not sure if I'm even on track of what the OP was asking.... but just my .02.

[besson3c]

Originally Posted by Zeeb 

Virtualization might be great for a company but not for my personal data. I've never liked the idea of paying someone a monthly fee for access to my personal files and I don't like the idea of taking it a step further with virtualization. While this fee might become wrapped up in a monthly broadband bill which I pay anyway I just don't trust it. This virtualization company will essentially "own" copies of my resume, my itunes music, my income tax returns, as well as the software and computing power I need to create new files-- and will be well within their rights to cut me off from access to all of that should I decide to terminate service. Not only would I not have access to my files, but I would not be able to create a new resume with the now worthless dead workstation screen.

What if you were able to buy an appliance for your household that allowed you to easily create VMs for the entire family, and did all of the other stuff we've been talking about (i.e. optimizing resource allocation)? Would you be cool with the idea then?

[Peter]

virtualization of servers, if that application that hosts all the virtualziation falls over, everything vanishes.
I don't get it.

[MacosNerd]

Originally Posted by besson3c 

What if you were able to buy an appliance for your household that allowed you to easily create VMs for the entire family, and did all of the other stuff we've been talking about (i.e. optimizing resource allocation)? Would you be cool with the idea then?

I wouldn't be, I'd rather have control and ownership of my data. I lose too much and gain too little to be tempted by this. There's little reason to create a VM for my family when I can have an actual computer for my family.

[besson3c]

Originally Posted by Peter 

virtualization of servers, if that application that hosts all the virtualziation falls over, everything vanishes.
I don't get it.

Only if you are serving your entire environment from a single piece of hardware, which is not a recommended setup for any server where it can be avoided.

In this event, if the admin was smart and was doing snapshots (i.e. backups) of his/her VMs, all they'd have to do is failover to this other machine. In many cases this would occur automatically, behind the scenes.

Contrast this to tethering hardware to the services that it runs on (as we have been doing), and you can see that the VM environment offers far better failsafes.

[MacosNerd]

Originally Posted by Peter 

virtualization of servers, if that application that hosts all the virtualziation falls over, everything vanishes.
I don't get it.

It makes a lot of sense for a company that has a large server farm. No need to have a large staff maintaining thousands of servers, they can consolidate to a small percentage. There's less benefit for doing this for the consumer especially given the privacy concerns that have been popping up in the news the past few years.

You touch on a point that if the VM goes, so doesn't your data. You need to trust the service has a great backup strategy in place. I'd rather be responsible for my own data and back it up myself. Just because a company says they have a comprehensive backup plan doesn't always mean they do.

[MacosNerd]

Originally Posted by besson3c 

In this event, if the admin was smart and was doing snapshots (i.e. backups) of his/her VMs, all they'd have to do is failover to this other machine. In many cases this would occur automatically, behind the scenes.

That's a mighty big if, what happens if he wasn't. You'd lose your data

[besson3c]

Originally Posted by MacosNerd 

That's a mighty big if, what happens if he wasn't. You'd lose your data

You'd lose your data with no backups on the servers that are run today without virtualization, and your service would be offline as well... What's your point?

[besson3c]

Originally Posted by BasketofPuppies 

I thought that Google offered an option to host Google Docs (and Google Apps) on your own servers. Maybe it was something planned for the future. And while running Google Docs on your own servers might not seem much different than running Microsoft Office over a network, do not discount the ease and convenience of Web browsers.

Regardless, of course Google is going to charge for this. Google is in business to make money, and advertising is inappropriate and distracting in a corporate environment.

I just don't see this...

The web browser is designed to not have direct access to your file system. We've seen the sort of damage that not counting on this can cause in ActiveX/IE. Applications like Final Cut Pro cache data locally for optimal performance. The application itself is designed for optimal performance. I don't see a practical way for a web browser to scale from running a simple word processor all the way up to an app like FCP, while maintaining this security model as we've come to know it today.

[besson3c]

Here are some other net gains from doing the virtualizing thing:

- less energy consumption across the entire power grid
- the ability to access your stuff from anywhere in the world
- no more having to buy a new computer every 3-5 years and making careful decisions about what to get
- easier to do things such as stream a signal to multiple displays (e.g. a DVD)
- no more having to install an OS and various apps when you can just pick from an existing image
- no more having to have particular hardware to run a particular OS (assuming that Apple eventually allows people to run OS X as a guest, and adjusts their product matrix and business models to accommodate these sorts of changes)