The Debian tag is there because they promoted the eglibc fork.
I am not quite sure that glibc is a GNU project. I thought for that to be true, the copyrights would have to be transferred to the FSF. Drepper posted a long rant a while back about how the FSF would do that over his cold, dead body.
When I wrote about pragmatism I was thinking of this problem where a modification to glibc's malloc() implementation broke the Adobe flash player. It is worth contrasting the attitude of Linus Torvalds in that thread with that of the glibc maintainers. I think most reasonable people would agree there is a trade off between supporting broken applications and ensuring things are done right. In this case, it would have cost glibc nothing to make a minor concession.
People who think tape is better because it's so cheap probably have not done their homework.
Reliability is a major factor. Tape is delicate. If you put your tape in storage how can you be sure that it will restore again correctly when you try to reuse it ? How can you be sure that whatever tape drive you're using by then will work ?
Disk backup is becoming common not simply because of cheap commodity disk drives, but because the software has improved so much as well, with technologies like ZFS and the similar technology included in the high end disk arrays from places like EMC or (especially) NetApp.
Deduplication keeps the data volumes down. You can't deduplicate tape backups, although some places sell a system which keeps deduplicated disk backups online and streams them out to tape offline.
With disks, you can RAID them together, with ZFS you even get triple parity RAID, and if you want to be really paranoid you can mirror that RAID array to another site. Tape can't do that.
With a decent array you can run regular scans and scrubs. You'll get early warning for failing disks, you can repair any corrupted sectors by reconstructing from the parity and you can rebuild entire disks onto hot spares when one fails. Tape can't do that either. And because of that resilience, and the random access capability, it is no longer necessary to do regular full backups, you only have to keep the incrementals. Typically, it usually works out - especially when combined with deduplication - that you need less than 1.8 times the amount of storage space to keep a month of backups.
When the disks get old and you can't replace them, you can migrate all the data and all the backup copies on them to a new array.
A decent pair of mirrored disk arrays not only acts as a backup, it can also act as a disaster recovery system. Since it can be a block-for-block mirror of your main disk array, if the building burns down and takes your main array with it, you just flick a switch on the backup array and it'll take over until you get a replacement.
There's something wrong with the the guy's claimed insurance costs. I put in a request on confused.com for a 2005 S2000GT, 5 years NCB, without a tracker, and the quote I got back was £1000. That's in Northern Ireland which has the highest average insurance in the country.
There's something very fishy about the claim made above. £8000/year is insane. Nobody would pay that.
You have a simple choice, which these guys are now attempting to execute - don't use GPL software, write your own.
If they think that they can produce software which is as capable as busybox is, and if they think they can attract OSS developers to contribute to it for free while seeing their copyrighted work locked away, more power to their elbow.
I really doubt that they'll be able to do it but it would be interesting to be proven wrong.
I agree that this is the way it is now. I have seen some projects done in India that were spectacularly bad. There is one particular case I am aware of where the development costs dropped dramatically when the project was returned onshore, and the quality of the deliverables went up significantly.
I have found that when an offshored project is going badly it tends to be because there is no adequate supervision of its progress. Business is good enough for outsourcing in India and China that they don't care too much about the state of the project if you're not asking about it. On the other hand, projects tend to go well if there are skilled, technical folks at home checking in on it constantly and keeping tabs on the progress. The savings, of course, tend not to be as dramatic as is promised, but they are nonetheless tangible and therefore difficult to resist for the bean counters.
To believe that this situation will remain static is naive. Back in the 70s, cars and consumer electronics kit from Japan, and later Korea, used to be the pits. Now these two countries are world leaders and almost all electronic goods come from there. Aside from a few recent mishaps, Japanese cars are known for being solid, fuel-efficient and they have the warranty to back it - around here Kia are doing a 7 year (!!) warranty. The day will eventually come - it may be a decade or two (but no more) before it does - when the second world economies have learned these skills and are upping the quality. Increasingly, IT engineers in the West will find themselves restricted to bespoke, specialist engineering work rather than full-blown projects.
It's wrong, of course, to believe that Indian engineers are all bad. It does draw my attention on the Usenet groups and message boards when I see the (amusing) message from "Amit" or "Srini" along the lines of "I have been tasked with porting the Linux kernel to an XYZ evaluation board. Please provide me with all documentation necessary to accomplish this task" or "please advise on the best approach to develop a new OLTP system supporting 1000 TPS". On the other hand, I've worked with some outstanding Indian engineers onshore. And of course, the best server-side wrapper for the best revision control system (gitolite) is developed by a TCS engineer - and a top notch job it is too.
Note, however, that using NFS as a datastore is a perfectly valid way to do things. VMware fully supports it, indeed alongside NetApp they encourage it - likewise Citrix on XenServer. NFS actually makes a lot of things easier. For example, you can browse right into the datastore and get at the logs, data files and disk images directly without having to mount whatever filesystem it is first. And for a small environment, there is absolutely no need for 10GBe or heavyweight FC. AoE is definitely risky though.
I would, on the other hand, say that it is nuts to use a hypervisor on a single box with a single RAID array. Virtualization is about putting a lot of eggs in one basket, so you want that basket to be distributed and redundant.
In terms of the main article itself.. Hyper-V and XenServer flatly do not compete in any kind of sense with VMware. I don't work for them but their product is impressive and it has come a long way. For example, VMW does memory compression and page sharing as well as ballooning, which so far neither Citrix nor Microsoft are doing, and these make it a lot easier to scale. This is they brought in the vmem tax - they're so good at being tight with memory that it saves you real cash. You can mostly seamlessly move VMs between the vSphere platform and your desktop running VMware Player - you can't do that with Hyper-V or Xen, not without some translation and conversion in between, which is never seamless.
There are things that are certainly frustrating about VMware. The licensing is esoteric and extremely complicated. It would be a lot nicer to have the core hypervisor and then purchase bolt-on licenses for the functionality you need; and they really should have done the VRAM thing that way. The things that are in different editions are weird. The distributed switch thing is a fantastic feature but why is it reserved for the highest end variation ? If I want it I have to pay for a shitload of stuff I don't need. Likewise, why do I have to have automated load distribution and automated power control before I can properly pool the resources of my servers ?
I believe the day will eventually come when a shrinkwrapped FOSS solution for doing virtualization with many of VMware's features becomes available with support and management tools that are as easy to use as VMware's. Xen does not quite do it for me as it kind of sits off to one side and the only people who are putting serious effort into it are Citrix. When it happens, it will probably be built on KVM, with mature tools and management components on top. Red Hat are heading in this direction and I would frankly not be massively surprised if Red Hat were to launch a standalone hypervisor product to compete directly with Xen and VMware.
Slightly more optimal if you do the first steps in reverse, ie
- install the OS first and do a cleanup on it (eg ccleaner on Windows), uninstall any crap that's on it, clear out logs/tmp files etc. - within the OS, write a big file full of zeros to fill the drive - ccleaner can do this easily by zeroing any unused space - dd will work if you can mount it from Linux.
Then use dd to capture the image as you suggested.
I am not sure the backwards compatibility argument completely stands up these days.
Back in the Amiga days, when the 68060 came out (we are going back the guts of 15 years here), the new processor dropped a few rarely-used instructions. To compensate, Motorola shipped a small library which allowed the old instructions to be simulated when they were detected via an illegal instruction trap.
By working with OS and compiler vendors, Intel could very easily deprecate and phase out all the old backwards-compatible instructions and addressing modes ahead of time. The only group of customers who would be effected by this would be folks who run old, unpatchable operating systems or software but yet also want to run the latest hardware. It's very hard for me to believe that this group is a significant %, especially not relative to the number of customers who are ready to patch their system and who want the benefits of the faster CPU.
It's interesting when you consider that none of the IRA's demands (ie the withdrawal of Britain from Ireland and the establishment of a 32-county Irish republic) were ever met. In fact, British withdrawal is probably further away now than it was when the IRA in its present guise got started in 1969.
I'm wondering what you think the IRA's "modus operandi" is. This is an organization that build napalm-like incendiary bombs and set them off in hotels, restaurants and pubs where civilians gathered in large numbers. I don't see why you think they would hesitate to attack a nuclear power station or other such facility.
I think that the accusation in question is far-fetched, although not impossible, but the important part is that the government are open to this accusation and there is no way for them to defend themselves against it. This undermines the whole democratic process.
It reminds me of a time when we were holding a controversial referendum here in Northern Ireland. One of the parties said that the government would try to throw the outcome of the referendum. The government responded by inviting that party to place their own seals on all the ballot boxes. After the poll was closed and when counting was about to commence, that party was invited to check that all of their seals were intact before the ballot boxes were reopened. As is typical with elections in the UK and Ireland, all of the parties were permitted to send their own delegations to monitor the counting process.
Explain to me how you accomplish that with electronic voting ?
It's called a pencil and a printed piece of paper with the candidates on.
This is the system used in the UK and a lot of other countries in the world. It can't be hacked, it is fully human readable, and it is completely transparent so any attempts to hack it immediately become obvious.
The election results are typically known beyond doubt within less than 24 hours of the poll closing, and the final results are typically declared within a day or two.
Reminds me of that old joke/urban myth where the USA spent $millions developing a pen that could be used in outer space, and the Soviets used a pencil.
As Barzini said in "The Godfather", "we are not communists".
Everything in our economy is about money, one way or another. It's money that decides in the first place what a "useful tool" is. And a good blacksmith who cares about his craft will command a higher price for his services than a poor one. It's in the blacksmith's interest to do a good job.
The mechanics of the software engineering business are somewhat different. It's been known for decades that duck tape programming delivers short term results quickly, but expensive and painful maintenance costs later on. But time to market is everything (indeed that is Microsoft's chief advantage). When you have deployed any kind of software system it's far from straightforward to switch to another vendor, in the way that you can switch to another blacksmith.
BTW by and large, Linux programmers - the major ones - are not about making tools, but they are about making the enhancements to the kernel that their employers pay them to do. Take away Red Hat, IBM, Novell and others, and what have you got ? Just because Linux is freely available, and volunteers can contribute to it, does not mean that it does not owe a significant proportion of its success to commercial concerns.
Who says Ballmer hadn't though about it ? Why put out a statement informing the world that you have spent lots of time thinking about it, and possibly reveal the fact that you're afraid ?
The reality in any organization is that there are good programmers and not-so-good programmers. And from time to time, even the good ones make mistakes. Different programmers have different strengths and weaknesses. That is why programming languages have things like type checking, and why software developers employ principles like encapsulation and data hiding. Your argument is that these practices are "restricting" clever programmers by making the software inflexible.
Taking your argument to its logical conclusion, you might say it isn't necessary to add debugging information or logging to programs if you higher decent programmers who never make mistakes.
What on earth is the point in creating this stupendous bandwidth in the access network when the core networks of countries across the internet are incapable of handling it ? This is like the ill-fated megahertz race in desktop PCs.
That assumes, of course, that protectionism protects your ability to get a job, which according to the prevailing economic thought, it doesn't. The problem here is not that foreign workers are cheap, it's that local workers are too expensive. You need to look at your cost of living; you probably spend one-third to one-half of your income on a mortgage or rent (and associated costs), and in the US, another significant proportion on family health insurance.
World War 2 spiralled out of a huge recession followed by the erection of protectionist barriers all across the world. History never repeats itself exactly, but we need to be careful here.
It is all moot anyway, as in 20 years time, the people who grew up pirating music will be in Government.
The people in government right now are the ones who had tape-to-tape decks and made their own compilations of their favourite albums for friends. Don't believe that hypocrisy isn't possible.
The USSR, having different needs and different mindsets, may have come up with unique technologies that where not tried here.
Well, they didn't. They used espionage to steal equipment from the West, which they then reverse engineered and duplicated. If there are any cases of actual Soviet technological innovation in the IT area, they're not very well documented.
Slashdot Mirror
The Debian tag is there because they promoted the eglibc fork.
I am not quite sure that glibc is a GNU project. I thought for that to be true, the copyrights would have to be transferred to the FSF. Drepper posted a long rant a while back about how the FSF would do that over his cold, dead body.
Just because they say it's not a fork doesn't mean it isn't a fork :)
When I wrote about pragmatism I was thinking of this problem where a modification to glibc's malloc() implementation broke the Adobe flash player. It is worth contrasting the attitude of Linus Torvalds in that thread with that of the glibc maintainers. I think most reasonable people would agree there is a trade off between supporting broken applications and ensuring things are done right. In this case, it would have cost glibc nothing to make a minor concession.
People who think tape is better because it's so cheap probably have not done their homework.
Reliability is a major factor. Tape is delicate. If you put your tape in storage how can you be sure that it will restore again correctly when you try to reuse it ? How can you be sure that whatever tape drive you're using by then will work ?
Disk backup is becoming common not simply because of cheap commodity disk drives, but because the software has improved so much as well, with technologies like ZFS and the similar technology included in the high end disk arrays from places like EMC or (especially) NetApp.
Deduplication keeps the data volumes down. You can't deduplicate tape backups, although some places sell a system which keeps deduplicated disk backups online and streams them out to tape offline.
With disks, you can RAID them together, with ZFS you even get triple parity RAID, and if you want to be really paranoid you can mirror that RAID array to another site. Tape can't do that.
With a decent array you can run regular scans and scrubs. You'll get early warning for failing disks, you can repair any corrupted sectors by reconstructing from the parity and you can rebuild entire disks onto hot spares when one fails. Tape can't do that either. And because of that resilience, and the random access capability, it is no longer necessary to do regular full backups, you only have to keep the incrementals. Typically, it usually works out - especially when combined with deduplication - that you need less than 1.8 times the amount of storage space to keep a month of backups.
When the disks get old and you can't replace them, you can migrate all the data and all the backup copies on them to a new array.
A decent pair of mirrored disk arrays not only acts as a backup, it can also act as a disaster recovery system. Since it can be a block-for-block mirror of your main disk array, if the building burns down and takes your main array with it, you just flick a switch on the backup array and it'll take over until you get a replacement.
There's something wrong with the the guy's claimed insurance costs. I put in a request on confused.com for a 2005 S2000GT, 5 years NCB, without a tracker, and the quote I got back was £1000. That's in Northern Ireland which has the highest average insurance in the country.
There's something very fishy about the claim made above. £8000/year is insane. Nobody would pay that.
No, the GPL does not force others to share.
You have a simple choice, which these guys are now attempting to execute - don't use GPL software, write your own.
If they think that they can produce software which is as capable as busybox is, and if they think they can attract OSS developers to contribute to it for free while seeing their copyrighted work locked away, more power to their elbow.
I really doubt that they'll be able to do it but it would be interesting to be proven wrong.
I agree that this is the way it is now. I have seen some projects done in India that were spectacularly bad. There is one particular case I am aware of where the development costs dropped dramatically when the project was returned onshore, and the quality of the deliverables went up significantly.
I have found that when an offshored project is going badly it tends to be because there is no adequate supervision of its progress. Business is good enough for outsourcing in India and China that they don't care too much about the state of the project if you're not asking about it. On the other hand, projects tend to go well if there are skilled, technical folks at home checking in on it constantly and keeping tabs on the progress. The savings, of course, tend not to be as dramatic as is promised, but they are nonetheless tangible and therefore difficult to resist for the bean counters.
To believe that this situation will remain static is naive. Back in the 70s, cars and consumer electronics kit from Japan, and later Korea, used to be the pits. Now these two countries are world leaders and almost all electronic goods come from there. Aside from a few recent mishaps, Japanese cars are known for being solid, fuel-efficient and they have the warranty to back it - around here Kia are doing a 7 year (!!) warranty. The day will eventually come - it may be a decade or two (but no more) before it does - when the second world economies have learned these skills and are upping the quality. Increasingly, IT engineers in the West will find themselves restricted to bespoke, specialist engineering work rather than full-blown projects.
It's wrong, of course, to believe that Indian engineers are all bad. It does draw my attention on the Usenet groups and message boards when I see the (amusing) message from "Amit" or "Srini" along the lines of "I have been tasked with porting the Linux kernel to an XYZ evaluation board. Please provide me with all documentation necessary to accomplish this task" or "please advise on the best approach to develop a new OLTP system supporting 1000 TPS". On the other hand, I've worked with some outstanding Indian engineers onshore. And of course, the best server-side wrapper for the best revision control system (gitolite) is developed by a TCS engineer - and a top notch job it is too.
Agreed with most of this.
Note, however, that using NFS as a datastore is a perfectly valid way to do things. VMware fully supports it, indeed alongside NetApp they encourage it - likewise Citrix on XenServer. NFS actually makes a lot of things easier. For example, you can browse right into the datastore and get at the logs, data files and disk images directly without having to mount whatever filesystem it is first. And for a small environment, there is absolutely no need for 10GBe or heavyweight FC. AoE is definitely risky though.
I would, on the other hand, say that it is nuts to use a hypervisor on a single box with a single RAID array. Virtualization is about putting a lot of eggs in one basket, so you want that basket to be distributed and redundant.
In terms of the main article itself .. Hyper-V and XenServer flatly do not compete in any kind of sense with VMware. I don't work for them but their product is impressive and it has come a long way. For example, VMW does memory compression and page sharing as well as ballooning, which so far neither Citrix nor Microsoft are doing, and these make it a lot easier to scale. This is they brought in the vmem tax - they're so good at being tight with memory that it saves you real cash. You can mostly seamlessly move VMs between the vSphere platform and your desktop running VMware Player - you can't do that with Hyper-V or Xen, not without some translation and conversion in between, which is never seamless.
There are things that are certainly frustrating about VMware. The licensing is esoteric and extremely complicated. It would be a lot nicer to have the core hypervisor and then purchase bolt-on licenses for the functionality you need; and they really should have done the VRAM thing that way. The things that are in different editions are weird. The distributed switch thing is a fantastic feature but why is it reserved for the highest end variation ? If I want it I have to pay for a shitload of stuff I don't need. Likewise, why do I have to have automated load distribution and automated power control before I can properly pool the resources of my servers ?
I believe the day will eventually come when a shrinkwrapped FOSS solution for doing virtualization with many of VMware's features becomes available with support and management tools that are as easy to use as VMware's. Xen does not quite do it for me as it kind of sits off to one side and the only people who are putting serious effort into it are Citrix. When it happens, it will probably be built on KVM, with mature tools and management components on top. Red Hat are heading in this direction and I would frankly not be massively surprised if Red Hat were to launch a standalone hypervisor product to compete directly with Xen and VMware.
Slightly more optimal if you do the first steps in reverse, ie
- install the OS first and do a cleanup on it (eg ccleaner on Windows), uninstall any crap that's on it, clear out logs/tmp files etc.
- within the OS, write a big file full of zeros to fill the drive - ccleaner can do this easily by zeroing any unused space - dd will work if you can mount it from Linux.
Then use dd to capture the image as you suggested.
I am not sure the backwards compatibility argument completely stands up these days.
Back in the Amiga days, when the 68060 came out (we are going back the guts of 15 years here), the new processor dropped a few rarely-used instructions. To compensate, Motorola shipped a small library which allowed the old instructions to be simulated when they were detected via an illegal instruction trap.
By working with OS and compiler vendors, Intel could very easily deprecate and phase out all the old backwards-compatible instructions and addressing modes ahead of time. The only group of customers who would be effected by this would be folks who run old, unpatchable operating systems or software but yet also want to run the latest hardware. It's very hard for me to believe that this group is a significant %, especially not relative to the number of customers who are ready to patch their system and who want the benefits of the faster CPU.
It's interesting when you consider that none of the IRA's demands (ie the withdrawal of Britain from Ireland and the establishment of a 32-county Irish republic) were ever met. In fact, British withdrawal is probably further away now than it was when the IRA in its present guise got started in 1969.
I'm wondering what you think the IRA's "modus operandi" is. This is an organization that build napalm-like incendiary bombs and set them off in hotels, restaurants and pubs where civilians gathered in large numbers. I don't see why you think they would hesitate to attack a nuclear power station or other such facility.
I think that the accusation in question is far-fetched, although not impossible, but the important part is that the government are open to this accusation and there is no way for them to defend themselves against it. This undermines the whole democratic process.
It reminds me of a time when we were holding a controversial referendum here in Northern Ireland. One of the parties said that the government would try to throw the outcome of the referendum. The government responded by inviting that party to place their own seals on all the ballot boxes. After the poll was closed and when counting was about to commence, that party was invited to check that all of their seals were intact before the ballot boxes were reopened. As is typical with elections in the UK and Ireland, all of the parties were permitted to send their own delegations to monitor the counting process.
Explain to me how you accomplish that with electronic voting ?
It's called a pencil and a printed piece of paper with the candidates on.
This is the system used in the UK and a lot of other countries in the world. It can't be hacked, it is fully human readable, and it is completely transparent so any attempts to hack it immediately become obvious.
The election results are typically known beyond doubt within less than 24 hours of the poll closing, and the final results are typically declared within a day or two.
Reminds me of that old joke/urban myth where the USA spent $millions developing a pen that could be used in outer space, and the Soviets used a pencil.
Are you going to keep them a secret too ?
In fairness, they're not the first to try this. IBM have done the same vis AIX. Peace, love and Linux .. but use AIX if you want to do real work.
As Barzini said in "The Godfather", "we are not communists".
Everything in our economy is about money, one way or another. It's money that decides in the first place what a "useful tool" is. And a good blacksmith who cares about his craft will command a higher price for his services than a poor one. It's in the blacksmith's interest to do a good job.
The mechanics of the software engineering business are somewhat different. It's been known for decades that duck tape programming delivers short term results quickly, but expensive and painful maintenance costs later on. But time to market is everything (indeed that is Microsoft's chief advantage). When you have deployed any kind of software system it's far from straightforward to switch to another vendor, in the way that you can switch to another blacksmith.
BTW by and large, Linux programmers - the major ones - are not about making tools, but they are about making the enhancements to the kernel that their employers pay them to do. Take away Red Hat, IBM, Novell and others, and what have you got ? Just because Linux is freely available, and volunteers can contribute to it, does not mean that it does not owe a significant proportion of its success to commercial concerns.
Who says Ballmer hadn't though about it ? Why put out a statement informing the world that you have spent lots of time thinking about it, and possibly reveal the fact that you're afraid ?
Dread ? Why ?
Does the USA dread Europe which, as a combined economy, is larger ?
Or do you just not like the Chinese
The reality in any organization is that there are good programmers and not-so-good programmers. And from time to time, even the good ones make mistakes. Different programmers have different strengths and weaknesses. That is why programming languages have things like type checking, and why software developers employ principles like encapsulation and data hiding. Your argument is that these practices are "restricting" clever programmers by making the software inflexible.
Taking your argument to its logical conclusion, you might say it isn't necessary to add debugging information or logging to programs if you higher decent programmers who never make mistakes.
We are supposed to be a Republic, not a pure Democracy
Your tone suggests that democracy isn't a very good idea.
What on earth is the point in creating this stupendous bandwidth in the access network when the core networks of countries across the internet are incapable of handling it ? This is like the ill-fated megahertz race in desktop PCs.
That assumes, of course, that protectionism protects your ability to get a job, which according to the prevailing economic thought, it doesn't. The problem here is not that foreign workers are cheap, it's that local workers are too expensive. You need to look at your cost of living; you probably spend one-third to one-half of your income on a mortgage or rent (and associated costs), and in the US, another significant proportion on family health insurance.
World War 2 spiralled out of a huge recession followed by the erection of protectionist barriers all across the world. History never repeats itself exactly, but we need to be careful here.
It is all moot anyway, as in 20 years time, the people who grew up pirating music will be in Government.
The people in government right now are the ones who had tape-to-tape decks and made their own compilations of their favourite albums for friends. Don't believe that hypocrisy isn't possible.
What happened to the scientific prowess of the Soviet Union?
The Soviets had scientific prowess ?
What did they do, other than steal Western innovations and pour massive state resources into enhancing them ?
The USSR, having different needs and different mindsets, may have come up with unique technologies that where not tried here.
Well, they didn't. They used espionage to steal equipment from the West, which they then reverse engineered and duplicated. If there are any cases of actual Soviet technological innovation in the IT area, they're not very well documented.