I think you're about 245K out with your cost estimate for the Sun box. Or you could even run Red Hat on a Sun x86 box and have Sun do the Red Hat support for you.
The point is that most organisations want to concentrate on running their business, not their IT.
Very few of the larger organisations I work with could care less about changing the source. The companies that supply their IT services might do... The point is that the OSS development model encourages fast changes and improvements, but the vast majority of companies are not interested in being developers themselves - they want finished products from someone else. They don't want to become coders and fiddlers, they want someone else to take care of that.
A medium sized company is not going to take SquirrelMail and fiddle with it to make it work on their platform, but they might get someone else to do it.
I like and use Linux, but these kinds of argumnents miss the point about what most IT departments are interested in from their technolody.
Linux' development model creates some good products, but few people want to actively change those products - they want someone else to do it for them.
Few Government departments are interested in looking at the code. They want systems that work. What use is say, a housing benefit system that's crap but has source code, when compared to a working closed source application. The cost of people who can look at code is also prohibitive. Even if you had them what are they going to do, look at the code and say 'yes it's crap and yes it'll cost you lots of money and time to hire someone to make it better'?
Lufthansa are nothing to do with the German Govt.
As you mention, support needs to be bought, so what's to stop those organisations buying off govt officials? Having the blueprints to the trucks your local council buys doesn't stop people being bought off, nor would source code being available for applications when services are being sold around those apps.
You can choose from 100s of companies for support of plenty of operating systems and solutions, Linux is nothing special. You aren't going to adjust the source code to your Red Hat dsitribution, you'd wait till Red Hat fixed it.
Linux isn't an open standard, but it does use open standards, as do many other OSes. Using open standards also doesn't necessarily makes something cheaper to maintain.
No government uses only one vendor for critical services.
Not really. Large companies aren't interested in creating custom distributions... Why bother making your own when you can 'outsource' the process to Red Hat, for example?
What's the feeling internally about the move from Solaris as the base database development platform to Linux? In technical terms Solaris and Sparc still have advantages over Intel and RAC runs well on Solaris.
I get the feeling that the Linux drive is more a business decision driven from the top. What's the feeling from the guys actually writing the products?
Solaris has been the development platform for the database for a long time. Solaris/Sparc still offer a lot of things Linux can't in technical terms and Oracle RAC, a cornerstone of Oracle sales reps' comp plans runs extremely well on Sun.
Changing the base development platform is a big move.
The Linux decision seems mainly to be a strategic business move driven from the top.
I'd be very interested to know how Oracle's developers feel towards the Linux move. They're the guys who really know the technical advantages between the various platforms Oracle runs on.
Except they announced nothing of the sort. Sun announced that they were going with 'Industry Standard' Linux distributions on the X86 servers they are shipping as customers weren't really interested in the Sun specific distribution that was 'Sun Linux 5.0'. On the Intel based kit they now sell you can buy and get support for Red Hat directly from Sun, or you can go with Solaris x86 and obviously get support for that too. Otherwise you are free to put whatever you like on the boxes.
Nice to see such a balanced, well thought out post. Sun's share price will probably crash tonight with the news you won't be using Java in the future. Good luck with your plumbing career.
I can understand about not wanting to pay exorbitant licensing fees, but his point was about paying money to foreign companies, with the implication that none of that money benefitted the local economy, which is plainly not the case.
So are you saying that Australia has noone capable of supporting Microsoft software? That's like saying that buying a PC in Australias just provides money to whichever foreign company made it.
All of the products and services sold have distribution channels and each part takes a cut, ploughing money back into the local economy.
Do Microsoft Australia and their employees pay no taxes? Who do you think gets commission for selling Microsoft products?
You don't have to fly in someone from the States every time you need something that isn't produced and locally.
I appreciate what you're saying, but it sounds pretty narrow minded. You must know that both are Unix type platforms and run all the GNU and OS apps and tools, so why discount Sun entirely when your skills are easily portable? Surely you'd be curious to work out what the best tool for the job would be?
The distinction the author's making is quite subtle, which might be why so many people aren't getting it here.
The premise is that Linux/Open Source advocates vociferously claim that Linux is both free as in beer and free is in 'libre', whilst the truth is always somewhat less straightforward. It might cost nothing, but has cost implications to run and maintain. The source might be 'free', but you have to buy 'real' apps to do a lot of useful stuff (Oracle for example). You may have the source, but frankly don't care - you don't insist on the blueprints for your car when you buy it... And so on and on and on.
What about predictable release schedules, excellent service, bundled apps and good ISV support? (All right, forget about the last bit - for now at least).
There's also the advantage of having your techies 'fluent' on the same OS throughout the datacentre, with one partner to deal with when things go well... or go wrong.
Linux/Sparc might be an option for small web serving, email and other stuff using OS components, but otherwise what app are you running? Your campus clustered SAP installaton with DR site across the other side of the world supporting your entire business is unlikely to be running Linux/Sparc.
Have you actually checked the prices for Sun's Linux boxes?
You can't just automatically replace any 4 way server with 4 one or two way Intels. There's a tonne of other stuff to consider. What app are you running? Set ups such as HPC/render farms will certainly be better off with a load of single cpu Xeons. Big app servers or databases needing a lot of memory or/and decent internal system bandwidth will benefit from a multi cpu machine, which is where Sparc performs very well. A decent workgroup Oracle server will perform very well on the v480. An Oracle RAC set up won't function well on lots of single or even dual cpu Intel machines and will cost a lot to manage.
With all of these systems it's horses for courses - there's a place for small Linux/Solaris x86/Intel machines and a place for the larger Sparc based boxes.
It depends. The 210/240 run Solaris for Sparc, which is a great OS. The internal system bandwidth is higher than the Dell and you get 4 Gbe NICs as standard, plus it'll go up to 8Gb of RAM.
I agree. Sun have Sun Rays or Sun workstations on everyone's desk, running CDE, used happily by all levels of staff, but it doesn't mean that CDE's ready to replace Windows on the desktop!
So the business benefit is that you have a load of source code that only an expensive consultant can manage? What's the business benefit in that? The benefit applies to a tiny minority of companies.
I think the point is the vendor supplying the support - Sun, for example, have docs.sun.com, where all Sun related stuff is neatly arranged. They have SunSolve which gives you some extra stuff too - it's neater than 'buy some O'Reilly books and go on the web and ask someone.'
If you had gold support from Sun you'd stick it on your server or mission critical workstations. It'd cost about 10-15% of the list price and importantly is a hardware AND software warranty, so you get support for the lot. If you had 800 highend Sun workstations you'd probably upgrade to silver support and have gold on the server. This would be important principally for hardware reasons, so you wouldn't have the money for your open source team. And if you had a problem, Sun would work on solving it.
Equally, if you did have an open source team, you'd need to make sure that team was doing stuff inline with all the other hospitals around the country or all the work you were doing would essentially be useless. It's also doubtful you could keep guys employed and motivated who had the skills to troubleshoot the entire gamut of software you were using. They'd neet to be able to do everything from configure Apache to write device drivers.
What you do is have a coherent IT strategy where you pick some key vendors who will partner well with you and you get your accontability and support from this relationship.
...and the boss says 'This will cost how much? I have to hire new people? Who is accountable? Why don't I just pick something off the shelf? Even if the cost over three years is greater, I have clear costings, accountability and support from the company that makes the software, which makes planning and accounting easier. Why are XYX corp going out of business? Even if they do, someone will buy their customer base?'.
It's hard enough to find good general IT staff, let alone competent programmers. Most businesses don't want to get anywhere near software development - they don't make their own cars, vans or lorries for God's sake, they buy them. Same goes for software.
The environment I work in is so conservative that we have customers who need convincing that major established vendors can meet warranty commitments on standard hardware. Suggesting to them that having the source code for their applications solves all their support problems is crazy. Source code is utterly irrelevant to them.
Your argument would come across as a sneery techie answer to these types of questions. You need to talk to management about solutions, rather than individual components. 'Mr Boss, this is how this works, this is the business benefit that we bring to the table. The components used are x,y,z and these are the routes we've evaluated regarding support. We believe we can meet SLAs based on the resources out there and based on the experience of other companies out there'
I honestly believe that 99% of busiensses out there really don't care about source code - they want something that works, integrates, scales, etc. They don't want to get into the sofware development business.
Just because Google use Intel and BSD or whatever for their type of app, doesn't mean the poster can miraculously replace their big iron with some cheap PCs running OSS. It's such a bland statement that gets made on Slashdot all the time with no real justification.
Slashdot Mirror
I think you're about 245K out with your cost estimate for the Sun box. Or you could even run Red Hat on a Sun x86 box and have Sun do the Red Hat support for you.
Why not just hire those cheaper coders for your OSS app from India as well?
I'm not talking about home use.
The point is that most organisations want to concentrate on running their business, not their IT.
Very few of the larger organisations I work with could care less about changing the source. The companies that supply their IT services might do... The point is that the OSS development model encourages fast changes and improvements, but the vast majority of companies are not interested in being developers themselves - they want finished products from someone else. They don't want to become coders and fiddlers, they want someone else to take care of that.
A medium sized company is not going to take SquirrelMail and fiddle with it to make it work on their platform, but they might get someone else to do it.
I like and use Linux, but these kinds of argumnents miss the point about what most IT departments are interested in from their technolody.
Linux' development model creates some good products, but few people want to actively change those products - they want someone else to do it for them.
Few Government departments are interested in looking at the code. They want systems that work. What use is say, a housing benefit system that's crap but has source code, when compared to a working closed source application. The cost of people who can look at code is also prohibitive. Even if you had them what are they going to do, look at the code and say 'yes it's crap and yes it'll cost you lots of money and time to hire someone to make it better'?
Lufthansa are nothing to do with the German Govt.
As you mention, support needs to be bought, so what's to stop those organisations buying off govt officials? Having the blueprints to the trucks your local council buys doesn't stop people being bought off, nor would source code being available for applications when services are being sold around those apps.
You can choose from 100s of companies for support of plenty of operating systems and solutions, Linux is nothing special. You aren't going to adjust the source code to your Red Hat dsitribution, you'd wait till Red Hat fixed it.
Linux isn't an open standard, but it does use open standards, as do many other OSes. Using open standards also doesn't necessarily makes something cheaper to maintain.
No government uses only one vendor for critical services.
Not really. Large companies aren't interested in creating custom distributions... Why bother making your own when you can 'outsource' the process to Red Hat, for example?
What's the feeling internally about the move from Solaris as the base database development platform to Linux? In technical terms Solaris and Sparc still have advantages over Intel and RAC runs well on Solaris.
I get the feeling that the Linux drive is more a business decision driven from the top. What's the feeling from the guys actually writing the products?
Solaris has been the development platform for the database for a long time. Solaris/Sparc still offer a lot of things Linux can't in technical terms and Oracle RAC, a cornerstone of Oracle sales reps' comp plans runs extremely well on Sun.
Changing the base development platform is a big move.
The Linux decision seems mainly to be a strategic business move driven from the top.
I'd be very interested to know how Oracle's developers feel towards the Linux move. They're the guys who really know the technical advantages between the various platforms Oracle runs on.
Except they announced nothing of the sort. Sun announced that they were going with 'Industry Standard' Linux distributions on the X86 servers they are shipping as customers weren't really interested in the Sun specific distribution that was 'Sun Linux 5.0'. On the Intel based kit they now sell you can buy and get support for Red Hat directly from Sun, or you can go with Solaris x86 and obviously get support for that too. Otherwise you are free to put whatever you like on the boxes.
Nice to see such a balanced, well thought out post. Sun's share price will probably crash tonight with the news you won't be using Java in the future. Good luck with your plumbing career.
I can understand about not wanting to pay exorbitant licensing fees, but his point was about paying money to foreign companies, with the implication that none of that money benefitted the local economy, which is plainly not the case.
So are you saying that Australia has noone capable of supporting Microsoft software? That's like saying that buying a PC in Australias just provides money to whichever foreign company made it.
All of the products and services sold have distribution channels and each part takes a cut, ploughing money back into the local economy.
Do Microsoft Australia and their employees pay no taxes? Who do you think gets commission for selling Microsoft products?
You don't have to fly in someone from the States every time you need something that isn't produced and locally.
I appreciate what you're saying, but it sounds pretty narrow minded. You must know that both are Unix type platforms and run all the GNU and OS apps and tools, so why discount Sun entirely when your skills are easily portable? Surely you'd be curious to work out what the best tool for the job would be?
You're actually allowed to install StarOffice on up to 5 other machines that you use!
What you actually need is a feature which automatically switches the phone off when it's in the cinema or theatre!
And have you checked the prices for the new Xeon based machines?
There's also the point of the bundled software and the one source for software and hardware support.
The distinction the author's making is quite subtle, which might be why so many people aren't getting it here.
The premise is that Linux/Open Source advocates vociferously claim that Linux is both free as in beer and free is in 'libre', whilst the truth is always somewhat less straightforward. It might cost nothing, but has cost implications to run and maintain. The source might be 'free', but you have to buy 'real' apps to do a lot of useful stuff (Oracle for example). You may have the source, but frankly don't care - you don't insist on the blueprints for your car when you buy it... And so on and on and on.
What about predictable release schedules, excellent service, bundled apps and good ISV support? (All right, forget about the last bit - for now at least).
There's also the advantage of having your techies 'fluent' on the same OS throughout the datacentre, with one partner to deal with when things go well... or go wrong.
Linux/Sparc might be an option for small web serving, email and other stuff using OS components, but otherwise what app are you running? Your campus clustered SAP installaton with DR site across the other side of the world supporting your entire business is unlikely to be running Linux/Sparc.
Have you actually checked the prices for Sun's Linux boxes?
You can't just automatically replace any 4 way server with 4 one or two way Intels. There's a tonne of other stuff to consider. What app are you running? Set ups such as HPC/render farms will certainly be better off with a load of single cpu Xeons. Big app servers or databases needing a lot of memory or/and decent internal system bandwidth will benefit from a multi cpu machine, which is where Sparc performs very well. A decent workgroup Oracle server will perform very well on the v480. An Oracle RAC set up won't function well on lots of single or even dual cpu Intel machines and will cost a lot to manage.
With all of these systems it's horses for courses - there's a place for small Linux/Solaris x86/Intel machines and a place for the larger Sparc based boxes.
It depends. The 210/240 run Solaris for Sparc, which is a great OS. The internal system bandwidth is higher than the Dell and you get 4 Gbe NICs as standard, plus it'll go up to 8Gb of RAM.
I agree. Sun have Sun Rays or Sun workstations on everyone's desk, running CDE, used happily by all levels of staff, but it doesn't mean that CDE's ready to replace Windows on the desktop!
So the business benefit is that you have a load of source code that only an expensive consultant can manage? What's the business benefit in that? The benefit applies to a tiny minority of companies.
I think the point is the vendor supplying the support - Sun, for example, have docs.sun.com, where all Sun related stuff is neatly arranged. They have SunSolve which gives you some extra stuff too - it's neater than 'buy some O'Reilly books and go on the web and ask someone.'
If you had gold support from Sun you'd stick it on your server or mission critical workstations. It'd cost about 10-15% of the list price and importantly is a hardware AND software warranty, so you get support for the lot. If you had 800 highend Sun workstations you'd probably upgrade to silver support and have gold on the server. This would be important principally for hardware reasons, so you wouldn't have the money for your open source team. And if you had a problem, Sun would work on solving it.
Equally, if you did have an open source team, you'd need to make sure that team was doing stuff inline with all the other hospitals around the country or all the work you were doing would essentially be useless. It's also doubtful you could keep guys employed and motivated who had the skills to troubleshoot the entire gamut of software you were using. They'd neet to be able to do everything from configure Apache to write device drivers.
What you do is have a coherent IT strategy where you pick some key vendors who will partner well with you and you get your accontability and support from this relationship.
...and the boss says 'This will cost how much? I have to hire new people? Who is accountable? Why don't I just pick something off the shelf? Even if the cost over three years is greater, I have clear costings, accountability and support from the company that makes the software, which makes planning and accounting easier. Why are XYX corp going out of business? Even if they do, someone will buy their customer base?'.
It's hard enough to find good general IT staff, let alone competent programmers. Most businesses don't want to get anywhere near software development - they don't make their own cars, vans or lorries for God's sake, they buy them. Same goes for software.
The environment I work in is so conservative that we have customers who need convincing that major established vendors can meet warranty commitments on standard hardware. Suggesting to them that having the source code for their applications solves all their support problems is crazy. Source code is utterly irrelevant to them.
Your argument would come across as a sneery techie answer to these types of questions. You need to talk to management about solutions, rather than individual components. 'Mr Boss, this is how this works, this is the business benefit that we bring to the table. The components used are x,y,z and these are the routes we've evaluated regarding support. We believe we can meet SLAs based on the resources out there and based on the experience of other companies out there'
I honestly believe that 99% of busiensses out there really don't care about source code - they want something that works, integrates, scales, etc. They don't want to get into the sofware development business.
Just because Google use Intel and BSD or whatever for their type of app, doesn't mean the poster can miraculously replace their big iron with some cheap PCs running OSS. It's such a bland statement that gets made on Slashdot all the time with no real justification.