I believe one day something like that will happen. There's no reason why anyone should have monopoly on providing enterprise-grade Linux and right now RH and SuSE do (and they also hold monopoly for enterprise Linux services since companies must buy their support).
And there's imbalance between ISVs and commercial Linux distributions - I believe many commercial software providers dislike Linux OS vendors because whatever gets popular (like J2EE) soon appears in GPL edition and bundled with OS, but they have to work with them anyway (or else).
(I don't have anything against OS-bundling OSS equivalents of commercial software but in a way it is unfair competition - if an app sells for 500 bucks and requires enterprise Linux (say 500 bucks) which already provides a similar albeit slightly inferior package, it gets harder for an ISV to sell their product/service*. If the ISV can sell their product with Debian Enterprise Linux, the customer may think twice before they buy Red Hat... The way it works now even JBoss which is open source has to compete with Red Hat, which I find odd.) * One cannot assume that Red Hat's packaging and QA efforts are more benefitial/important than the application vendor's coding and support (we can even assume that the ISVs gives away the source and sells support/maintenance).
And finally, with many servers being used in application-specific way, I wonder why would enterprises buy generic OS support which contributes so little to what they actually need (application support). Only reading support conditions for enterprise Linux makes me sick - there's nothing they are obliged to do (like, their responsibility is that "/etc/init.d/smb start" exits with an "OK" (big deal!), but you can't ask anything about smb.conf, etc. )
With really free enterprise Linux it would be open for all and Linux OS vendors would finally feel the heat...
>If you're a big business you're likely going to want support. If you're not a big business, then postgres will likely suit your needs, and maybe even suit them better than ingres.
If you're not a big business then - what the hell! - they're the same both are free and you can't afford to pay for support anyway.
I think their search technology is mediocre - just try searching for anything really popular and you'll see what you get - links to search engine spammers....
Google's cluster: that's a possibility although I don't know (I don't think they know either) what exactly they can do with their cluster at the moment. Before I've thought about these: a) enterprise mail hosting No - it's very competitive and they don't have many features (for example the folders - any MTA would see everything in INBOX) b) enterprise search Well they do that already... How many enterprises would let them rather than IBM search internal databases (not just Intranet)? c) high performance computing - not feasible since their servers are already loaded to a reasonable level.. d) a new generation search - something new such as smart searching of multiple languages with English summaries, image search, etc. This could be cool but there's no reason to believe Google can do that better than any of existing companies already specialized in those fields.
They're too dependent on ads and the field is very very competitive. I don't think they'll be able to sustain this share price.
>I just see sun as trying to use linux in selfish kind of way, not help or foster it
Gee, what else is any company that owns a commercial distribution supposed to do? If they buy SuSE (and Novell), they should use it and abuse it as much as legally and comercially possible.
IBM stated back in Y2000 that they'll never enter the Linux distribution business and that hasn't changed - the reason is that it doesn't make (as much) money.
"Unfortunately, the Xserve lacks redundant fans or power supplies, and it lets you create redundant hard drives only via the feature-poor and unreliable software-RAID implementation included in Apple's Disk Utility. When we tested our Xserve with two 60GB drives formatted as a mirrored array, Disk Utility didn't let us mirror a drive on which we already had data, forcing us to erase all drives in the RAID set and start over. (Our advice is to decide on your setup before you begin to configure your server.) "
Source: http://www.macworld.com/2002/11/reviews /xserve/ (True, 2002 but I also heard that from a customer three months ago)
"Despite rebuilding my 10.2-created RAID from the command-line (which took 5.5 hours for 130GB of data on a 2x250GB drive!), I was still getting flaky performance, out-of-sync errors and no rebuild option under 10.3.4"
Source: http://www.versiontracker.com/dyn/morei nfo/macosx/ 20768&mode=feedback&vid=All (A July 2004 post).
Also, I've seen it in action and it had mediocre performance with stability problems similar to what they say on the Web.
>With price/GB 3x cheaper than Dell (and even less vs. IBM, HP and Sun), and certified Linux+Windows compatibility, I'd say Xserve is a pretty interesting product.
Gezuz, cheaper than Dell's what? This is a mid-end department-level FC-SATA disk array! What EMC (Dell), IBM, HP and Sun sell are real enterprise disk arrays for building SANs. It's a completely different class of products. Well, perhaps you don't have experience with storage... The only thing interesting about this product is that Apple's marketing fooled people into believing this is some kind of EMC killer.
Better compare it with Adaptec FS4500 (starts at $6K without the disks, has active-active controllers (unlike Apple), up to 1GB cache per controller (Apple's got up to 512MB I think)) and RAID software probably works.
Apple has no experience with enterprise storage (which is obvious from the continous RAID software problems) and that's it. It's ludicrous to buy storage from such company.
if that is on one location, that isn't cost effective and convenient - seven RAIDs and seven servers (and seven tape drives, seven backups to monitor)...
>I'll just have to bring on my next trick: using standard hard drives as cache for a massive array of RAID tape drives!
Yeah such software exists already - the viewpoint is the opposite from yours (the software moves infrequently used data from (expensive) disks to (cheap) tapes and back when neccessary) but the effect is the same - most frequently accessed data and/or most recently created data (depending on the settings) is on HDD, old shit is on the tapes. To the end user, it's all transparent - you see all files on the file server but physically they might not be there but instead on off-line (tapes) or near-line (el cheapo disk arrays) storage.
With their disk array, when a disk fails you can't rebuild it online - you need to do it offline. Also, if one channel fails, the other can't take over, so that's another way to downtime.
If one cares about RAID and redundancy/availability, they shouldn't use Apple's product in the first place.
My thinking wasn't in that direction - more like yearly CPU upgrades until the box can be retired. That's would be useful for commercial applications where licensing is per-processor based - Oracle and such. One reboot and you've got a 50% faster CPU...
Gee the article says "But they have proven "exceptionally difficult to program" and problematic at certain performance levels," (quote from a 2004 study by the President's High-End Computing Revitalization Task Force.)
Now you come up with some MPI theory negating a study by HPC experts.. WTF? Anyway, your MPI coding skills are not a relevant to the main issue raised in the article.
>If I know MPI (which I do), I can code on any cluster out there, and it will perform similarly, with a linear time dependance on network speed.
Exactly - with a linear time dependEnce on network speed. Which means that for some tasks, as explained in the article, MPI-type clusters are not the right solution. They are the cheapest solution - or merely a "workaround".
And finally, any task with huge data sets that cannot be split in subsets doesn't lend itself to MPI-type processing.
The article's point is that instead of evolutionary progress (replacing Myrinet with Infiniband or Xeons with Opterons), the government should spend more to create revolutionary improvement.
>The socket shuffle is no big deal, since by the time you want to upgrade your CPU you most likely will want a new motherboard too for new memory or other technologies. Motherboards aren't all that expensive regardless...
It is a big deal - one will be able to upgrade a Xeon system to Itanium with minimum effort. Of course, the OS and the apps will have to be re-installed, but the box itself won't require rewiring and rack-mounting. In data center environment the mobo itself doesn't contribute to performance a lot, but the key is in lowering the TCO, so it will matter. And as the article said, having interchangeable mobos will make every Intel Xeon reseller also an Intel Itanium reseller at no extra cost.
The very fact that "mobos aren't all that expensive regardless" is the reason for this move. What is expensive is fscking with small volume of specialized parts, seinding technicians and engineers to Itanium training, etc.
Of course, I don't know exactly, I imagine they could figure out the precision, get the idea how the system works, etc. and it'd be easier to avoid it or create a counterweapon. In any case, I primarily wanted to say that the idea that source is open should be also considered from the counter-intelligence standpoint.
>What stops me from considering that this is "dumping" is the result, a company can block a competing piece of work because of some arbitrarily determined amount of money that should be charged.
Again, I agree with you. Sometime last week in another post I argued how OSS isn't dumping because there's no fair way to determine and compare the cost.
> Scary thought there.
Still, imagine the government decides OSS is unfair competition (consider that no OSS developers pay taxes although customers profit from using OSS; I somewhere read how the government prefers commercial software because they can collect sales tax on it) and that all OSS must be paid for:-)
This is a new idea (to me) - they could decide minimum selling prices for workstation-grade Linux and for server-grade OSS to provide a level playing field for commercial software. Then the money collected could be used for financing development of OSS via government grants.
This is just a random thought, but who knows, maybe that would be better than buying OSS from commercial OS distributors (I have always wondered why governments don't standardize on Debian - if only one government chipped in 1-2m dollars a year - and that is less than they pay now for enterprise Linux - they could have Debian certified for major enterprise hardware and software).... Oh, well..
I agree with you; still, it's a very complex issue - on and off, I've been thinking about "economics of OSS" for a while now and although I agree with the general idea that ISVs undercut by OSS can switch to writing apps for which there are no OSS equivalents, it can't work because software is easy to re-write and oftentimes the value isn't in the amount or quality of code but in the idea behind it. It takes a long time to think up a product, write it, document it, etc. and it doesn't take as long to reverse engineer and rewrite it.
If one looks at some very popular OSS applications he'll see that many of their features are recycled and rewritten from proprietary software. At the same time OSS community is against software patents (and not only software patents).
If software, ideas and processes remain patentable, OSS and proprietary software can co-exist. I think a lot depends on outcome of changes to patent laws.
P.S. I just thought about this - what happens if a company wants to ruin their competitor who has single product - they start giving the software away for free (since they can still make some money on other products). Now the single-product company goes bust. Without OSS that would have been called dumping. Would that be legal with OSS?
>..is why the Microsoft-supplied OpenGL screen saver on my Compaq PC (installed by Compaq, no less) keeps crashing? You know, the one gives the date and time in a cool spinny font.
I believe RH used to have the same feature in version 6.x - screen saver would by default be on, so if one installed X Windows, the sucker would sometimes freeze the system or start eating 99% of CPU.
In your case perhaps it's graphics card driver or its onboard RAM... Or power saving mode (if it's on)? Why do you run that screen saver anwyay?
Slashdot Mirror
Me too.
Last night I checked out several AMD64 boards - as they all used VIA chipsets, I decided to get a new Xeon64-32 board instead.
I've had enough of low quality chipsets (such as SiS video problems on Linux and VIA driver problems on Windows).
I believe one day something like that will happen.
There's no reason why anyone should have monopoly on providing enterprise-grade Linux and right now RH and SuSE do (and they also hold monopoly for enterprise Linux services since companies must buy their support).
And there's imbalance between ISVs and commercial Linux distributions - I believe many commercial software providers dislike Linux OS vendors because whatever gets popular (like J2EE) soon appears in GPL edition and bundled with OS, but they have to work with them anyway (or else).
(I don't have anything against OS-bundling OSS equivalents of commercial software but in a way it is unfair competition - if an app sells for 500 bucks and requires enterprise Linux (say 500 bucks) which already provides a similar albeit slightly inferior package, it gets harder for an ISV to sell their product/service*. If the ISV can sell their product with Debian Enterprise Linux, the customer may think twice before they buy Red Hat... The way it works now even JBoss which is open source has to compete with Red Hat, which I find odd.)
* One cannot assume that Red Hat's packaging and QA efforts are more benefitial/important than the application vendor's coding and support (we can even assume that the ISVs gives away the source and sells support/maintenance).
And finally, with many servers being used in application-specific way, I wonder why would enterprises buy generic OS support which contributes so little to what they actually need (application support). Only reading support conditions for enterprise Linux makes me sick - there's nothing they are obliged to do (like, their responsibility is that "/etc/init.d/smb start" exits with an "OK" (big deal!), but you can't ask anything about smb.conf, etc. )
With really free enterprise Linux it would be open for all and Linux OS vendors would finally feel the heat...
>If you're a big business you're likely going to want support. If you're not a big business, then postgres will likely suit your needs, and maybe even suit them better than ingres.
If you're not a big business then - what the hell! - they're the same both are free and you can't afford to pay for support anyway.
Seems that PostreSQL loses in both cases?
They should have thought about that long time ago, in which case they would have figured it out and paid Debian to get all enterprise certifications.
Thanks for the useless insight of yours.
25m shares sold on open market at say 100 bucks per share is $2.5b.
In other words to buy this stuff back Goog will have to write a biiiig check in order to get those shares to the people they were promised to.
Then if they manage to pull that off, the existing (other) shareholders can class-sue someone in the management for gross negligence.
I think their search technology is mediocre - just try searching for anything really popular and you'll see what you get - links to search engine spammers....
Google's cluster: that's a possibility although I don't know (I don't think they know either) what exactly they can do with their cluster at the moment.
Before I've thought about these:
a) enterprise mail hosting
No - it's very competitive and they don't have many features (for example the folders - any MTA would see everything in INBOX)
b) enterprise search
Well they do that already... How many enterprises would let them rather than IBM search internal databases (not just Intranet)?
c) high performance computing - not feasible since their servers are already loaded to a reasonable level..
d) a new generation search - something new such as smart searching of multiple languages with English summaries, image search, etc. This could be cool but there's no reason to believe Google can do that better than any of existing companies already specialized in those fields.
They're too dependent on ads and the field is very very competitive. I don't think they'll be able to sustain this share price.
>I just see sun as trying to use linux in selfish kind of way, not help or foster it
Gee, what else is any company that owns a commercial distribution supposed to do?
If they buy SuSE (and Novell), they should use it and abuse it as much as legally and comercially possible.
Really?
IBM stated back in Y2000 that they'll never enter the Linux distribution business and that hasn't changed - the reason is that it doesn't make (as much) money.
IBM's CEO said in 2000 that IBM doesn't want to be in the OS business and that hasn't changed.
The reason of course is that the OS doesn't make money.
>For accurate information, check the source.
s /xserve/
i nfo/macosx/ 20768&mode=feedback&vid=All
c _FS4500_hard_drive_ array/4014-3033_9-30801460.html?tag=pl&q=Adaptec%2 C+Inc.s upport/techspecs .jsp?sess=no&language=English+US&prodkey=FS4500&ca t=%2FProduct%2FFS4500
Screw the source.
"Unfortunately, the Xserve lacks redundant fans or power supplies, and it lets you create redundant hard drives only via the feature-poor and unreliable software-RAID implementation included in Apple's Disk Utility. When we tested our Xserve with two 60GB drives formatted as a mirrored array, Disk Utility didn't let us mirror a drive on which we already had data, forcing us to erase all drives in the RAID set and start over. (Our advice is to decide on your setup before you begin to configure your server.) "
Source:
http://www.macworld.com/2002/11/review
(True, 2002 but I also heard that from a customer three months ago)
"Despite rebuilding my 10.2-created RAID from the command-line (which took 5.5 hours for 130GB of data on a 2x250GB drive!), I was still getting flaky performance, out-of-sync errors and no rebuild option under 10.3.4"
Source:
http://www.versiontracker.com/dyn/more
(A July 2004 post).
Also, I've seen it in action and it had mediocre performance with stability problems similar to what they say on the Web.
>With price/GB 3x cheaper than Dell (and even less vs. IBM, HP and Sun), and certified Linux+Windows compatibility, I'd say Xserve is a pretty interesting product.
Gezuz, cheaper than Dell's what?
This is a mid-end department-level FC-SATA disk array!
What EMC (Dell), IBM, HP and Sun sell are real enterprise disk arrays for building SANs.
It's a completely different class of products.
Well, perhaps you don't have experience with storage...
The only thing interesting about this product is that Apple's marketing fooled people into believing this is some kind of EMC killer.
Better compare it with Adaptec FS4500 (starts at $6K without the disks, has active-active controllers (unlike Apple), up to 1GB cache per controller (Apple's got up to 512MB I think)) and RAID software probably works.
Source:
Price:
http://shopper.cnet.com/Adapte
Specs:
http://www.adaptec.com/worldwide/
Apple has no experience with enterprise storage (which is obvious from the continous RAID software problems) and that's it.
It's ludicrous to buy storage from such company.
>As of 6.0SP2 (due out soon, hopefully) you can no longer create a window without a statusbar.
I can't wait for that - with the current version I can't make the goddamn thing _display_ a status bar after restart....
if that is on one location, that isn't cost effective and convenient - seven RAIDs and seven servers (and seven tape drives, seven backups to monitor)...
>I'll just have to bring on my next trick: using standard hard drives as cache for a massive array of RAID tape drives!
Yeah such software exists already - the viewpoint is the opposite from yours (the software moves infrequently used data from (expensive) disks to (cheap) tapes and back when neccessary) but the effect is the same - most frequently accessed data and/or most recently created data (depending on the settings) is on HDD, old shit is on the tapes.
To the end user, it's all transparent - you see all files on the file server but physically they might not be there but instead on off-line (tapes) or near-line (el cheapo disk arrays) storage.
With their disk array, when a disk fails you can't rebuild it online - you need to do it offline. Also, if one channel fails, the other can't take over, so that's another way to downtime.
If one cares about RAID and redundancy/availability, they shouldn't use Apple's product in the first place.
As many have said - you can get the same thing for less from other SATA raid vendors.
Secondly this storage does NOT perform as well. I have seen other similarly priced disk arrays get up to 30% better performance.
And as a matter of fact, that guy just needs 4 SATA disks and LVM.
My thinking wasn't in that direction - more like yearly CPU upgrades until the box can be retired.
That's would be useful for commercial applications where licensing is per-processor based - Oracle and such. One reboot and you've got a 50% faster CPU...
Gee the article says "But they have proven "exceptionally difficult to program" and problematic at certain performance levels," (quote from a 2004 study by the President's High-End Computing Revitalization Task Force.)
Now you come up with some MPI theory negating a study by HPC experts.. WTF?
Anyway, your MPI coding skills are not a relevant to the main issue raised in the article.
>If I know MPI (which I do), I can code on any cluster out there, and it will perform similarly, with a linear time dependance on network speed.
Exactly - with a linear time dependEnce on network speed. Which means that for some tasks, as explained in the article, MPI-type clusters are not the right solution. They are the cheapest solution - or merely a "workaround".
And finally, any task with huge data sets that cannot be split in subsets doesn't lend itself to MPI-type processing.
The article's point is that instead of evolutionary progress (replacing Myrinet with Infiniband or Xeons with Opterons), the government should spend more to create revolutionary improvement.
Could you afford the electricity bill?
>The socket shuffle is no big deal, since by the time you want to upgrade your CPU you most likely will want a new motherboard too for new memory or other technologies. Motherboards aren't all that expensive regardless...
It is a big deal - one will be able to upgrade a Xeon system to Itanium with minimum effort.
Of course, the OS and the apps will have to be re-installed, but the box itself won't require rewiring and rack-mounting.
In data center environment the mobo itself doesn't contribute to performance a lot, but the key is in lowering the TCO, so it will matter.
And as the article said, having interchangeable mobos will make every Intel Xeon reseller also an Intel Itanium reseller at no extra cost.
The very fact that "mobos aren't all that expensive regardless" is the reason for this move. What is expensive is fscking with small volume of specialized parts, seinding technicians and engineers to Itanium training, etc.
Of course, I don't know exactly, I imagine they could figure out the precision, get the idea how the system works, etc. and it'd be easier to avoid it or create a counterweapon.
In any case, I primarily wanted to say that the idea that source is open should be also considered from the counter-intelligence standpoint.
>What stops me from considering that this is "dumping" is the result, a company can block a competing piece of work because of some arbitrarily determined amount of money that should be charged.
:-)
Again, I agree with you.
Sometime last week in another post I argued how OSS isn't dumping because there's no fair way to determine and compare the cost.
> Scary thought there.
Still, imagine the government decides OSS is unfair competition (consider that no OSS developers pay taxes although customers profit from using OSS; I somewhere read how the government prefers commercial software because they can collect sales tax on it) and that all OSS must be paid for
This is a new idea (to me) - they could decide minimum selling prices for workstation-grade Linux and for server-grade OSS to provide a level playing field for commercial software. Then the money collected could be used for financing development of OSS via government grants.
This is just a random thought, but who knows, maybe that would be better than buying OSS from commercial OS distributors (I have always wondered why governments don't standardize on Debian - if only one government chipped in 1-2m dollars a year - and that is less than they pay now for enterprise Linux - they could have Debian certified for major enterprise hardware and software).... Oh, well..
> then release the known-secure code to the community. .. then release the known-secure code to the terrorist community.
I agree with you; still, it's a very complex issue - on and off, I've been thinking about "economics of OSS" for a while now and although I agree with the general idea that ISVs undercut by OSS can switch to writing apps for which there are no OSS equivalents, it can't work because software is easy to re-write and oftentimes the value isn't in the amount or quality of code but in the idea behind it.
It takes a long time to think up a product, write it, document it, etc. and it doesn't take as long to reverse engineer and rewrite it.
If one looks at some very popular OSS applications he'll see that many of their features are recycled and rewritten from proprietary software.
At the same time OSS community is against software patents (and not only software patents).
If software, ideas and processes remain patentable, OSS and proprietary software can co-exist. I think a lot depends on outcome of changes to patent laws.
P.S. I just thought about this - what happens if a company wants to ruin their competitor who has single product - they start giving the software away for free (since they can still make some money on other products). Now the single-product company goes bust. Without OSS that would have been called dumping. Would that be legal with OSS?
> ..is why the Microsoft-supplied OpenGL screen saver on my Compaq PC (installed by Compaq, no less) keeps crashing? You know, the one gives the date and time in a cool spinny font.
I believe RH used to have the same feature in version 6.x - screen saver would by default be on, so if one installed X Windows, the sucker would sometimes freeze the system or start eating 99% of CPU.
In your case perhaps it's graphics card driver or its onboard RAM... Or power saving mode (if it's on)?
Why do you run that screen saver anwyay?
I used to say it as a student, then as married, then as divorced.
Then I found a new young girlfriend and now I have less money then I was a student...