You are talking about cloud for servers, and abstracting away things you want to do on your severs to cloud services like AWS. The article is talking about using SaaS / Cloud to replace desktop apps and thus allow you to replace desktops. Think something like Chrome OS. The hardware strategy for the server team is obviously not going to have any impact on the desktop staffing needs.
In terms of the desktop, I call what you are talking about in terms of VMWare or Citrix desktop virtualization not thin client and not cloud. It sounds like you have a nice VDI setup but that probably takes more staff than a bad Managed Desktop setup.
You are right that Apple's interface is very stable. Mainly because they pulled so far ahead with interface in the early 2000s. That being said the GUI has gotten way way better since 10.1.
Quartz Extreme Universal Access (this was huge for lots of people) Fast User Switching Expose Preview later: instant alpha, graphic extraction Quicktime integration Spotlight Dashboard Automator Integrated H.264/AVC Resource forks handled by command line XCode visual modeling, remote debugging, integrated reference library Integrated Dictionary Quartz Composer Webclip Stacks iChat integration Podcast Capture Spaces 3D dock
1) You want to run newer applications 2) You want to use newer hardware 3) You want to to run older applications on more demanding data 4) You want some of the collaboration features for phone / tablet...
Otherwise no reason to upgrade. I know a woman who still writes on xywrite with a dot matrix printer since everything is setup perfectly for her. When she's finished she hands a floppy off to an editor who translates it. Old computers still work. Ask the mainframers.
I don't think it is going to be that bad. Looking at your user ID, you are going to be able to handle issueing a command to the ROM to either turn off secure boot or use a new cert.
Huh? Active Desktop was awesome. Active X allowed for really power distributed applications that were web based. I'd love to have channels for my phone today. PointCast which is still the best screen saver I ever had kinda like news360.
I can't recommend or not recommend but http://www.koolspan.com/ offers a product to do this. Otherwise Nokia has been doing it for 8 years though with Symbian not Android.
Understood. But it is rarer than you think because they want to charge extra for renting the box and they can't if it ain't optional. Many of them also support Tivo in some other way.
Just because the server is running a linux hypervisor doesn't mean the resulting Windows virtual desktop is any different to manage
I think you are missing the point of cloud. There is no "Windows virtual desktop". This isn't fake thin client like citrix, this is the real thing. You don't run a normal desktop OS at all on the client either virtualized or actual.
Yeah, they're fully generic, as long as they're running Windows XP, a specific version of IE6, and have precisely the correct DLLs installed to handle the stupid "thin client browser-based application" the company had written during its thin client fetish circa 2001.
That's the point why I don't consider remote clients to be thin clients.
Your Java story is the problem of supporting bad software. Bad software is bad software. I will agree though that Java doesn't deliver on the "write once run anywhere" the way it should. HTML (not Javascript) is pretty good on working anywhere, Flash ain't 1/2 bad. But if you want to see this done right look at the old textmode: curses libraries, termcap and terminfo.
is a reduction in application management. Exactly. Though once you get rid of highly versatile desktop apps don't forget you pick up complexity in your server apps.
and the network guy who makes sure that LAN and WAN keep talking to one another
That BTW becomes vastly more complex. Traffic is now up 1000% at least. All sorts of issues that weren't worth worrying about are now worth worrying about.
Well thought out comments. I think the big savings that people forget is real estate savings. I suspect cloud will drive IT cost up substantially and unless this is paired with mobile / home / outsourcing it will die.
"We've been down this road before."
You address a bunch of technical problems that mostly didn't stop centralization in earlier generations by analogy. There are two main factors that drive the move towards distribution is the demand for customization and performance/cost. The more centralized a service is the more complex it is. The more complex it is the higher the cost of dealing with edge case features. As those edge case features aren't dealt with rogue alternate systems start getting used. That is how Microsoft beat the mainframes.
The second is that local data and cost. You see this already on mobile devices moving towards local apps and local storage. And the reason is that objective-c is at least 100x less CPU intensive than JavaScript for the same task. Once you move to dumb, all CPU activities are happening on the server and all data is flowing through the network. The edge cases no one thinks about suddenly become the bulk of your traffic and costs.
no more long-term planning of purchases of those servers
That depends if the company is moving to 100% off the shelf cloud solutions or not. If not, they are making long-term planning in funding development and signing long term service contracts.
the point is, that you don't have to move *everything* to the cloud in order to make savings and find other benefits.
Then you are just talking about what stuff goes on which server and of course there are benefits of AWS for some apps. To have what Gartner is talking about everything needs to be off the PC. Local machines to be disposable must be generic.
I would be surprised if 10% of all IT jobs are now done offshore
What! Think about all the application development jobs. Think about all the inhouse software development jobs using remote teams. Think about all those call center support jobs. Hardware manufacturing jobs. I'd be shocked if more than 25% of the total possible jobs are domestic.
Dumb terminals have been around for decades and didn't end the desktop.
First off in anything it was mostly the other way around. The rich desktops killed the dumb terminals. But... if you think about what happened right after, we moved towards the world of locked down desktops. The rich configurable desktops were killed by a push back in the dumb terminal direction.
Terminal servers were a terrible idea. Take desktop thick client apps and deploy them remotely. That makes no sense at all. You get the disadvantages of thick client hardware, the disadvantages of tremendously expensive servers. OTOH because the apps aren't distributed you don't get the advantages in versatility of thick client. The worst of all worlds.
In the browser based solution, the goal is not to save money on client hardware. You can't, because browser apps are something like 100x more CPU intensive than local applications you might very well up your client costs. The goal though is the clients are fully generic. Any and all sufficiently powerful computers can run the remote app without any additional support. That allows for work from home culture, which saves you on real estate.
Sure this cycle has been repeated since the 1960s.
1) There are real advantages to centralization for some applications 2) There are real advantages to distributed for some applications 3) There are substantial additional costs in being both distributed and centralized
3 encourages people to move towards one extreme or the other. The conflict between 1 and 2 pushes the back towards the center.
In terms of malware, Windows created a computer monoculture for the enterprise desktops. Servers have more diversity than desktops. You might see a large drop off in malware.
This was one of the early arguments for Unix/Linux.
Windows because it offers the possibility of a rich client has: complex breakable hardware, which is unique to the user, a complex OS and applications susceptible to malware. A thin client erases all those issues. The hardware itself is far less breakable, and isn't unique to the user. You can just have spares and have them fixed "whenever". The OS just has to boot the hardware and connect to the servers, and the applications all exist remotely. Think about your television as the hardware, the cable box as the OS and the shows as being applications. The TV rarely breaks and when it does it can replaced with another generic television.
Now.... you are replacing your desktop team with a more complex system admin and operations team because the local system But right now, as a legacy of Windows, most companies have both complex server solutions and complex desktop solutions.
If one can get to the command line you can execute "perl vurus.pl" and use the path to find the interpreter. That works wherever perl is installed as advised by most Perl packages.
I agree. But look what happened there. Because of diversity you moved from just running a script to having to install a script and do a workaround. Throw in 100 of those obstacles and suddenly malware gets too complicated.
Monocultures in software are an effect of the necessity of training or knowledge of software. A company can not reasonably expect a new employee to know a piece of esoteric office software that is only known by 5% of the people qualified for the job. Therefore the company would have to train the person on the the company's software and most new employees will waste time getting familiar with the local office suite....
That would apply equally to any other piece of software used by industry. Yet we have more than one CAD, more than one video editing app, more than one music editing app... And before Microsoft was so dominant office suites were getting more diverse not less, even though that same economics applied. Certainly business training costs are lower in a mono culture that was the big argument companies offered as they moved from a world of 60% Microsoft office to 90% Microsoft office during the early to mid 90s. But if mono cultures are the desired end then Linux will never be successful at all on the desktop since that breaks the mono culture.
There are costs to a mono culture for business in terms of lack of diversity. Companies have vastly different workflows and approval processes and Office (especially without a great Sharepoint setup) just doesn't support these workflows well. That's including things like inherent problems in WYSIWYG vs. WYSIWYM (a Linux norm).
In any case, I think if we are going to do this analysis it is only reasonable to assume a Linux world of diversity. Not one where Linux imitates the properties of Windows.
I do not see that Linux desktop users will be any different. If Linux gains business share there will be many more non-technical users which will be very different from today's geeky users.
Oh absolutely, no question. Your point about updates is correct, but your are missing my point that Linux updates and Windows updates are of a different type. There will be fewer.
1) A person could distribute a GPLed application themselves, redistribution would then be questionable but not neccesarily illegal. (This was precisely the situation with KDE when QT was under that license)
2) A person could distribute a BSD/MIT licensed code which depended on that library. They would just want to make sure the recipient understood that he's getting a conjoined work under 2 different licenses.
3) A person could distribute a closed source application using the QT style license as long as they didn't sell it.
Slashdot Mirror
Exactly. If ebooks and PoD weren't changing the buying patterns it wouldn't matter much. But they are making possible very low volume books...
You are talking about cloud for servers, and abstracting away things you want to do on your severs to cloud services like AWS. The article is talking about using SaaS / Cloud to replace desktop apps and thus allow you to replace desktops. Think something like Chrome OS. The hardware strategy for the server team is obviously not going to have any impact on the desktop staffing needs.
In terms of the desktop, I call what you are talking about in terms of VMWare or Citrix desktop virtualization not thin client and not cloud. It sounds like you have a nice VDI setup but that probably takes more staff than a bad Managed Desktop setup.
You are right that Apple's interface is very stable. Mainly because they pulled so far ahead with interface in the early 2000s. That being said the GUI has gotten way way better since 10.1.
Quartz Extreme
Universal Access (this was huge for lots of people)
Fast User Switching
Expose
Preview later: instant alpha, graphic extraction
Quicktime integration
Spotlight
Dashboard
Automator
Integrated H.264/AVC
Resource forks handled by command line
XCode visual modeling, remote debugging, integrated reference library
Integrated Dictionary
Quartz Composer
Webclip
Stacks
iChat integration
Podcast Capture
Spaces
3D dock
and that's just to 10.5.
1) You want to run newer applications
2) You want to use newer hardware
3) You want to to run older applications on more demanding data
4) You want some of the collaboration features for phone / tablet...
Otherwise no reason to upgrade. I know a woman who still writes on xywrite with a dot matrix printer since everything is setup perfectly for her. When she's finished she hands a floppy off to an editor who translates it. Old computers still work. Ask the mainframers.
I don't think it is going to be that bad. Looking at your user ID, you are going to be able to handle issueing a command to the ROM to either turn off secure boot or use a new cert.
Huh? Active Desktop was awesome. Active X allowed for really power distributed applications that were web based. I'd love to have channels for my phone today. PointCast which is still the best screen saver I ever had kinda like news360.
use encryption for texting and phone calls.
I can't recommend or not recommend but http://www.koolspan.com/ offers a product to do this. Otherwise Nokia has been doing it for 8 years though with Symbian not Android.
Understood. But it is rarer than you think because they want to charge extra for renting the box and they can't if it ain't optional. Many of them also support Tivo in some other way.
Just because the server is running a linux hypervisor doesn't mean the resulting Windows virtual desktop is any different to manage
I think you are missing the point of cloud. There is no "Windows virtual desktop". This isn't fake thin client like citrix, this is the real thing. You don't run a normal desktop OS at all on the client either virtualized or actual.
My cable box is a piece of shit, and I want to put a big gaping hole through it with a .12ga slug.
Maybe but how many times has the cable company had to service it? And couldn't they just mail you a new one?
As an aside. Have you considered going cable card and getting a Tivo? I've been a customer for years and love it.
Yeah, they're fully generic, as long as they're running Windows XP, a specific version of IE6, and have precisely the correct DLLs installed to handle the stupid "thin client browser-based application" the company had written during its thin client fetish circa 2001.
That's the point why I don't consider remote clients to be thin clients.
Your Java story is the problem of supporting bad software. Bad software is bad software. I will agree though that Java doesn't deliver on the "write once run anywhere" the way it should. HTML (not Javascript) is pretty good on working anywhere, Flash ain't 1/2 bad. But if you want to see this done right look at the old textmode: curses libraries, termcap and terminfo.
Glad you agree with the distinction.
is a reduction in application management.
Exactly. Though once you get rid of highly versatile desktop apps don't forget you pick up complexity in your server apps.
and the network guy who makes sure that LAN and WAN keep talking to one another
That BTW becomes vastly more complex. Traffic is now up 1000% at least. All sorts of issues that weren't worth worrying about are now worth worrying about.
Well thought out comments. I think the big savings that people forget is real estate savings. I suspect cloud will drive IT cost up substantially and unless this is paired with mobile / home / outsourcing it will die.
"We've been down this road before."
You address a bunch of technical problems that mostly didn't stop centralization in earlier generations by analogy. There are two main factors that drive the move towards distribution is the demand for customization and performance/cost. The more centralized a service is the more complex it is. The more complex it is the higher the cost of dealing with edge case features. As those edge case features aren't dealt with rogue alternate systems start getting used. That is how Microsoft beat the mainframes.
The second is that local data and cost. You see this already on mobile devices moving towards local apps and local storage. And the reason is that objective-c is at least 100x less CPU intensive than JavaScript for the same task. Once you move to dumb, all CPU activities are happening on the server and all data is flowing through the network. The edge cases no one thinks about suddenly become the bulk of your traffic and costs.
no more long-term planning of purchases of those servers
That depends if the company is moving to 100% off the shelf cloud solutions or not. If not, they are making long-term planning in funding development and signing long term service contracts.
the point is, that you don't have to move *everything* to the cloud in order to make savings and find other benefits.
Then you are just talking about what stuff goes on which server and of course there are benefits of AWS for some apps. To have what Gartner is talking about everything needs to be off the PC. Local machines to be disposable must be generic.
The CRM solution is already on a server running remotely from the desktop. Moving to dumb terminals doesn't eliminate server teams it increases them.
I would be surprised if 10% of all IT jobs are now done offshore
What! Think about all the application development jobs. Think about all the inhouse software development jobs using remote teams. Think about all those call center support jobs. Hardware manufacturing jobs. I'd be shocked if more than 25% of the total possible jobs are domestic.
Dumb terminals have been around for decades and didn't end the desktop.
First off in anything it was mostly the other way around. The rich desktops killed the dumb terminals. But... if you think about what happened right after, we moved towards the world of locked down desktops. The rich configurable desktops were killed by a push back in the dumb terminal direction.
Thanks for the tip!
Terminal servers were a terrible idea. Take desktop thick client apps and deploy them remotely. That makes no sense at all. You get the disadvantages of thick client hardware, the disadvantages of tremendously expensive servers. OTOH because the apps aren't distributed you don't get the advantages in versatility of thick client. The worst of all worlds.
In the browser based solution, the goal is not to save money on client hardware. You can't, because browser apps are something like 100x more CPU intensive than local applications you might very well up your client costs. The goal though is the clients are fully generic. Any and all sufficiently powerful computers can run the remote app without any additional support. That allows for work from home culture, which saves you on real estate.
Sure this cycle has been repeated since the 1960s.
1) There are real advantages to centralization for some applications
2) There are real advantages to distributed for some applications
3) There are substantial additional costs in being both distributed and centralized
3 encourages people to move towards one extreme or the other. The conflict between 1 and 2 pushes the back towards the center.
In terms of malware, Windows created a computer monoculture for the enterprise desktops. Servers have more diversity than desktops. You might see a large drop off in malware.
This was one of the early arguments for Unix/Linux.
Windows because it offers the possibility of a rich client has: complex breakable hardware, which is unique to the user, a complex OS and applications susceptible to malware. A thin client erases all those issues. The hardware itself is far less breakable, and isn't unique to the user. You can just have spares and have them fixed "whenever". The OS just has to boot the hardware and connect to the servers, and the applications all exist remotely. Think about your television as the hardware, the cable box as the OS and the shows as being applications. The TV rarely breaks and when it does it can replaced with another generic television.
Now .... you are replacing your desktop team with a more complex system admin and operations team because the local system But right now, as a legacy of Windows, most companies have both complex server solutions and complex desktop solutions.
If one can get to the command line you can execute "perl vurus.pl" and use the path to find the interpreter. That works wherever perl is installed as advised by most Perl packages.
I agree. But look what happened there. Because of diversity you moved from just running a script to having to install a script and do a workaround. Throw in 100 of those obstacles and suddenly malware gets too complicated.
Monocultures in software are an effect of the necessity of training or knowledge of software. A company can not reasonably expect a new employee to know a piece of esoteric office software that is only known by 5% of the people qualified for the job. Therefore the company would have to train the person on the the company's software and most new employees will waste time getting familiar with the local office suite. ...
That would apply equally to any other piece of software used by industry. Yet we have more than one CAD, more than one video editing app, more than one music editing app... And before Microsoft was so dominant office suites were getting more diverse not less, even though that same economics applied. Certainly business training costs are lower in a mono culture that was the big argument companies offered as they moved from a world of 60% Microsoft office to 90% Microsoft office during the early to mid 90s. But if mono cultures are the desired end then Linux will never be successful at all on the desktop since that breaks the mono culture.
There are costs to a mono culture for business in terms of lack of diversity. Companies have vastly different workflows and approval processes and Office (especially without a great Sharepoint setup) just doesn't support these workflows well. That's including things like inherent problems in WYSIWYG vs. WYSIWYM (a Linux norm).
In any case, I think if we are going to do this analysis it is only reasonable to assume a Linux world of diversity. Not one where Linux imitates the properties of Windows.
I do not see that Linux desktop users will be any different. If Linux gains business share there will be many more non-technical users which will be very different from today's geeky users.
Oh absolutely, no question. Your point about updates is correct, but your are missing my point that Linux updates and Windows updates are of a different type. There will be fewer.
That's not true.
1) A person could distribute a GPLed application themselves, redistribution would then be questionable but not neccesarily illegal. (This was precisely the situation with KDE when QT was under that license)
2) A person could distribute a BSD/MIT licensed code which depended on that library. They would just want to make sure the recipient understood that he's getting a conjoined work under 2 different licenses.
3) A person could distribute a closed source application using the QT style license as long as they didn't sell it.
I think by proprietary he meant commercial. I suggested http://pigale.sourceforge.net/license_Qt.html
Well yes there are licenses like that, they are called non commercial licenses. For example: http://pigale.sourceforge.net/license_Qt.html