Been using Linux SW raid in the 2.4 kernel series for a year+ now and it has worked like a champ, with both IDE and SCSI devices. All disk servers were SMP (overkill but management wanted it that way). Dunno what you screwed up.
If your criteria for an adequate disk server include either (a) high performance or (b) long-term maintainability, then you should choose SW raid.
Most HW raid systems, especially cheapo PCI cards, but even expensive Fibre Channel-SCSI3 rackmount monsters, offer either extremely primitive performance metrics or none at all. With SW raid, you normally get the full performance-monitoring and tuning capabilities of the host OS. Big win. You will also get better performance from a SW raid, given the same drive layout, and as long as you do not use the box for anything else at the same time. It should be obvious but some people don't believe this.
The other big win is more important when you spend more money than $3000 (a pittance in this market): there's no hardware manufacturer to get bought, go out of business, or change product lines. No multi-thousand-dollar support contract or custom software to configure the RAID or any of that other crap. Trust me, when your dedicated RAID box's motherboard flakes on you and you discover the manufacturer has gone out of business, you'll be cursing yourself for choosing HW raid every time you search Ebay for a replacement part.
Not to mention that commodity, general-purpose HW is always cheaper to replace, and its performance/price ratio grows much faster than special purpose HW. The HW raid system with the 200MHz i760 and 64MB RAM might have looked great in 2000 but now you're stuck with the proprietary on-disk format of an out-of-business vendor with no way out except to build a new system of the same capacity and copy everything over. (In the case of large data warehouses, "full backups" don't usually happen.)
HW raid was compelling in the past. Now, with commodity hardware so cheap, and open, stable SW raid systems floating around, you'd be a fool not to prefer them in many situations. If you want a fire-and-forget dedicated box, go for it. But be ready for the "forget" part in a year or two.
While it's true that MS was certainly within their rights to dismiss the guy, the issue here is whether or not they should have dismissed him.
No secrets were revealed, no damage was done; he offered to take the post down and was refused. There's something of a lack of proportionality between the offense and the penalty.
Just because it's legal doesn't make it right.
Also, as others have pointed out, the negative publicity MS is getting as a result is probably more than enough to offset whatever benefit they might have derived from sacking the guy.
Sun's in a lot of trouble, and a lot of it is because of their resistance to the proliferation of Linux. They also have an image problem and they sell (a lot of) underpowered, overpriced hardware. This should sound like mid-90s Apple, and the prescription for a fix is similar:
Get on the Linux bandwagon.
Development on Solaris is costing Sun a lot of money, and it's money they don't have to waste. Instead of spending R&D and support dollars on Solaris, they should spend that time and money on adding those few important technologies they have into Linux. Yes, this will put them under the GPL, but in return they get to vastly increase the size of their user base, get device support and performance improvements for free, and garner a lot of goodwill among users and developers.
Linux is probably just recently reached the maturity level (thanks in part to IBM, who figured this out a few years ago) where it could (with Sun's contributions) support their big hardware as well as Solaris does. The time to make the move is now.
(Apple example: they grabbed BSD and turned it into OS X. Brilliant move, as it saved them the waste of working on the whole OS and allowed them to leverage BSD's capabilities and concentrate on the parts that made OS X great.)
Cultivate the image as a premium hardware vendor.
Sun has no chance, I repeat, no chance of competing long-term with Dell (and others) on Dell's playing field. The biggest reason is that Dell doesn't do research! Sun's already behind in the cost-cutting game from the start, and they can't hope to compete when they cannot leverage economies of scale either (on the SPARC).
There is a place for the SPARC, and it would be a shame to see it go. Sun's big machines (E10K range) are fantastic machines and Intel/AMD don't have any interest in helping Sun build machines to compete with their big customers. Ditto for IBM and the Power chips.
Unfortunately, the biggest purchasing cuts when the economy is weak come from big-iron orders, as companies try to make do with what they already have or press less-capable hardware into use. This is a cruel fact that Sun's competitors in this area (HP, IBM) are dealing with too. IBM has its consulting dominance and HP its printing/imaging cash-cow to fall back on, though; Sun has neither of those. They need to come up with some sort of profit-making group to get them through the lean times. (This, they'll have to come up with on their own. Put those R&D dollars to work!)
Sun can make money selling smaller boxes if they, like Apple, position themselves correctly in the market. Small boxes suffer from an inexorable downward price pressure. They have to make that Sun logo on the case worth something. The stories about the Ultra 10s are apocryphal -- very nice, expensive workstations that last about 9 months before requiring complete overhauls. That cannot happen. Sun is only going to have success in the new workstation market if they make great products that provide some compelling value for purchasers that justifies their price premium. Right now, they have lost that (and not just because they ship with Solaris).
(Apple example: all of their hardware since the original iMac.)
Change their image.
Sun is perceived by many in the industry, both customers and peers, and being a bunch of arrogant pricks. All of their salespeople are clueless assholes; McNealy is a myopic bozo who thinks he's competing with Microsoft while IBM, HP, and Dell eat his lunch; they're running off all of the famous people who work there (never a good sign!). So goes the perception.
Along with the business-strategy changes listed above, they have to get people to think about Sun in a positive light again. Don't piss off your customers. In the late 90s Sun was rolling in dotcom cash from their big-ticket purchases and started to act like it was better than its customers. Now, they're starved for business. The vision at t
Ahh, a college student. He still thinks the fundamental aspect of an economy is based on the scarcity of goods.
What _is_ it based on, then, genius? If no good were scarce, there would be no need to work for anything. An economy is a system for distributing goods from those that have them to those that want or need them.
We realize that right now, wealth and prosperity is acquired by this thing we call "work". We can't have a society where there is no work, otherwise what the fuck is the point of living? The community is the reason we exist.
This sounds like a halfwit's Marxism. Are we talking about economics or psychology? Humans, like all other creatures, are capable of consuming more than they produce. (I'm talking about economics here, not food-goes-in-poo-goes-out conservation-of-matter stuff.) If you have some brilliant psychomagic that can make people produce more than they consume, I'd love to see it. History has shown us that the most efficent* way to get people to produce more is to dangle the carrot of more consumption in front of them. Oh wait, I'm talking economics again. Silly me! The Bolshevik utopia will be forever out of my reach.
The laws of supply and demand work for people too. Right now, the demand for human ingenuity and artistic vision is very very low, so the majority of people are corporate/government bureaucrats, impotent university drones, or modern day servants.
And this is different from history in what way? The vast majority of all people at any given time in human history have been engaged in drudgery. To think of some great human achievements: man on the moon? Takes an awful lot of nose-to-the-grindstone work to make a big-ass rocket. Pyramids? Takes slaves to drag all those rocks. And so on for millenia. The "ingenuity" and "artistic vision" gets all of the praise, but it takes more than ingenuity and vision to make things happen in this world. (Now I sound like the Marxist!)
The "cult of efficiency" of economics is interested in maximizing the amount of value that people get for their work. When it comes down to it, what's the fundamental value in the system? Human pleasure. The question of _whose_ pleasure it is, well, that's the sticky one, isn't it? It sounds like you think your pleasure is worth more than mine, or of whoever else might take your job. Thus you must be protected. Sounds like a brilliant (and morally justifiable) plan to me.
*efficient in the sense that you maximize, in whatever scale you are looking at, the values you...well, value. Other concerns (environment, "morality", etc.) might or might not be considered. Such is the nature of a maximization problem.
Maybe windows is the (sigh, spelling sucks) equivalent (did i get that right?) of an old hammer and the rest of us is waiting and working towards the pneumatic hammer?
You can spell "pneumatic" without a hitch but you have trouble with "equivalent"???
Linux's NFS server support has gotten leaps and bounds better since about 2.4.14 or so. The "bleeding edge" NFS stuff works quite well. Is it quite up to Sun's standard? No, it's not. But it's getting close.
Of course, the perception of Linux's NFS support was probably done a fair amount of harm by Red Hat's bastardized 2.4.18 kernel that shipped with 7.3. BROKEN NFS client support out of the box with anything but Linux servers. Sent our big Sun servers into the ozone every time the load grew beyond "trivial."
If you're interested in good NFS performance, throw your Red Hat kernel away and build a clean 2.4.20, or one with the NFS patches* if you're running servers.
There are darn few things which Sun is cost effective for anymore. Running a big DB, etc... But the word is Intel for file and print servers and smaller app servers as well. Choose a Linux box with commodity hardware and you could have that entry level blade for about $1,000. If you are worried about the reliability of commodity hardware, get a back up. You still saved half your money.
Exactly right, and this is why I think they are doing themselves serious harm by still pretending like there is profit to be made in their vertical strategy. Linux hasn't hurt Sun that badly in the computer room, but it is eating its lunch in the network room and on the desktop. Who in their right mind would shell out twice or thrice the bucks for a Sun desktop box when they can get functional equivalence or better with a Linux/Intel one?
Seems like I keep beating this horse...Sun is not going to be able to compete any more on the user end unless they join the commoditization parade. Period. They don't want to recognize this but they really need to. The no-alternative days that saw any corporation needing Unix run to one of the big vendors and forking over millions for end-to-end installations are long over. SGI has been down this path already; Sun won't be able to subsidize their desktop hardware with server revenue for much longer.
You also mention another reason I hate Solaris -- the dearth of what have become common tools and features for an OS. Yes, I know that kickback from Veritas is nice, but volume management support is about six months away from becoming a throw-in. Oh yes, it makes perfect business sense to maintain an entire separate toolchain for things like 'ls' and 'grep'. Etc.
And I really wish Sun would stop calling their shitty workstations "Blades", since that term has become accepted to mean something else entirely.
Sun's core market is not being stolen by Linux. Sun's big money comes from big servers with lots of processors. Their core market is the SunFire line (x800, and the E1xk), not the Blade line(comparable to PC).
True.
Sun's getting killed by the general reluctance of big companies to spend money on huge servers. This is in part because of general global instability (terrorism, economic downturn, imminent war, etc.), in part because there are lots of slightly-used servers floating around thanks to the dot-coms all going under, and in part because they have no idea what they are doing.
Sun is throwing money away pushing Solaris for x86 (if you can call this "pushing") and even for their workstation-class machines. Sun's throwing money away continuing to make workstation-class machines; the move away from Sun on the scientific desktop to Linux is well underway. The Blade I am stuck with is a fine machine, for a Sun, but a PC costing half as much with any recent Linux leaves it in the dust performance-wise, hardware-support-wise, and usability-wise.
Sun is in serious trouble if they don't stop wasting valuable resources on areas of business that will only become less profitable for them. I don't think they realize this, or at least they don't want to realize it.
There's a difference here: There have to be more drivers included with an x86 operating system distribution because there are so many different variations of x86 hardware, unlike with the SPARC platform where Sun Microsystems and SPARC Intl. are pretty much in control. Drivers take disc space. Granted, drivers won't fill a CD, and the Solaris OE has always lacked drivers for the newest hardware, but it's still something to be considered.
I bet SPARC is bigger and I'll tell you why -- sparcv9 is 64 bit. If Sun is still shipping 32-bit as well, then you've got the text equivalent of triple the size of the 32-bit sparc only. (I don't want to get into the relative SPARC/x86 code size war. My experience is that 32-bit sparc code is about 90% of the size of equivalent x86 code.)
I really don't think Sun is expecting to make any money on Solaris 9 anyway, so their $20 charge seems more like a disincentive for people to actually use it. Given their incredible marketing acumen and crack support team (sarcasm?), it wouldn't surprise me.
This isn't Insightful at all. Sun is not battling Microsoft in the consumer desktop market, nor does it have any reason to try. We're talking about servers here, and all MS servers* run on Intel hardware.
*with the possible exception of the last 3 machines on Earth still running NT on Alpha...
Slashdot Mirror
Been using Linux SW raid in the 2.4 kernel series for a year+ now and it has worked like a champ, with both IDE and SCSI devices. All disk servers were SMP (overkill but management wanted it that way). Dunno what you screwed up.
If your criteria for an adequate disk server include either (a) high performance or (b) long-term maintainability, then you should choose SW raid.
Most HW raid systems, especially cheapo PCI cards, but even expensive Fibre Channel-SCSI3 rackmount monsters, offer either extremely primitive performance metrics or none at all. With SW raid, you normally get the full performance-monitoring and tuning capabilities of the host OS. Big win. You will also get better performance from a SW raid, given the same drive layout, and as long as you do not use the box for anything else at the same time. It should be obvious but some people don't believe this.
The other big win is more important when you spend more money than $3000 (a pittance in this market): there's no hardware manufacturer to get bought, go out of business, or change product lines. No multi-thousand-dollar support contract or custom software to configure the RAID or any of that other crap. Trust me, when your dedicated RAID box's motherboard flakes on you and you discover the manufacturer has gone out of business, you'll be cursing yourself for choosing HW raid every time you search Ebay for a replacement part.
Not to mention that commodity, general-purpose HW is always cheaper to replace, and its performance/price ratio grows much faster than special purpose HW. The HW raid system with the 200MHz i760 and 64MB RAM might have looked great in 2000 but now you're stuck with the proprietary on-disk format of an out-of-business vendor with no way out except to build a new system of the same capacity and copy everything over. (In the case of large data warehouses, "full backups" don't usually happen.)
HW raid was compelling in the past. Now, with commodity hardware so cheap, and open, stable SW raid systems floating around, you'd be a fool not to prefer them in many situations. If you want a fire-and-forget dedicated box, go for it. But be ready for the "forget" part in a year or two.
While it's true that MS was certainly within their rights to dismiss the guy, the issue here is whether or not they should have dismissed him.
No secrets were revealed, no damage was done; he offered to take the post down and was refused. There's something of a lack of proportionality between the offense and the penalty.
Just because it's legal doesn't make it right.
Also, as others have pointed out, the negative publicity MS is getting as a result is probably more than enough to offset whatever benefit they might have derived from sacking the guy.
Sun's in a lot of trouble, and a lot of it is because of their resistance to the proliferation of Linux. They also have an image problem and they sell (a lot of) underpowered, overpriced hardware. This should sound like mid-90s Apple, and the prescription for a fix is similar:
Development on Solaris is costing Sun a lot of money, and it's money they don't have to waste. Instead of spending R&D and support dollars on Solaris, they should spend that time and money on adding those few important technologies they have into Linux. Yes, this will put them under the GPL, but in return they get to vastly increase the size of their user base, get device support and performance improvements for free, and garner a lot of goodwill among users and developers.
Linux is probably just recently reached the maturity level (thanks in part to IBM, who figured this out a few years ago) where it could (with Sun's contributions) support their big hardware as well as Solaris does. The time to make the move is now.
(Apple example: they grabbed BSD and turned it into OS X. Brilliant move, as it saved them the waste of working on the whole OS and allowed them to leverage BSD's capabilities and concentrate on the parts that made OS X great.)
Sun has no chance, I repeat, no chance of competing long-term with Dell (and others) on Dell's playing field. The biggest reason is that Dell doesn't do research! Sun's already behind in the cost-cutting game from the start, and they can't hope to compete when they cannot leverage economies of scale either (on the SPARC).
There is a place for the SPARC, and it would be a shame to see it go. Sun's big machines (E10K range) are fantastic machines and Intel/AMD don't have any interest in helping Sun build machines to compete with their big customers. Ditto for IBM and the Power chips.
Unfortunately, the biggest purchasing cuts when the economy is weak come from big-iron orders, as companies try to make do with what they already have or press less-capable hardware into use. This is a cruel fact that Sun's competitors in this area (HP, IBM) are dealing with too. IBM has its consulting dominance and HP its printing/imaging cash-cow to fall back on, though; Sun has neither of those. They need to come up with some sort of profit-making group to get them through the lean times. (This, they'll have to come up with on their own. Put those R&D dollars to work!)
Sun can make money selling smaller boxes if they, like Apple, position themselves correctly in the market. Small boxes suffer from an inexorable downward price pressure. They have to make that Sun logo on the case worth something. The stories about the Ultra 10s are apocryphal -- very nice, expensive workstations that last about 9 months before requiring complete overhauls. That cannot happen. Sun is only going to have success in the new workstation market if they make great products that provide some compelling value for purchasers that justifies their price premium. Right now, they have lost that (and not just because they ship with Solaris).
(Apple example: all of their hardware since the original iMac.)
Sun is perceived by many in the industry, both customers and peers, and being a bunch of arrogant pricks. All of their salespeople are clueless assholes; McNealy is a myopic bozo who thinks he's competing with Microsoft while IBM, HP, and Dell eat his lunch; they're running off all of the famous people who work there (never a good sign!). So goes the perception.
Along with the business-strategy changes listed above, they have to get people to think about Sun in a positive light again. Don't piss off your customers. In the late 90s Sun was rolling in dotcom cash from their big-ticket purchases and started to act like it was better than its customers. Now, they're starved for business. The vision at t
Ahh, a college student. He still thinks the fundamental aspect of an economy is based on the scarcity of goods.
What _is_ it based on, then, genius? If no good were scarce, there would be no need to work for anything. An economy is a system for distributing goods from those that have them to those that want or need them.
We realize that right now, wealth and prosperity is acquired by this thing we call "work". We can't have a society where there is no work, otherwise what the fuck is the point of living? The community is the reason we exist.
This sounds like a halfwit's Marxism. Are we talking about economics or psychology? Humans, like all other creatures, are capable of consuming more than they produce. (I'm talking about economics here, not food-goes-in-poo-goes-out conservation-of-matter stuff.) If you have some brilliant psychomagic that can make people produce more than they consume, I'd love to see it. History has shown us that the most efficent* way to get people to produce more is to dangle the carrot of more consumption in front of them. Oh wait, I'm talking economics again. Silly me! The Bolshevik utopia will be forever out of my reach.
The laws of supply and demand work for people too. Right now, the demand for human ingenuity and artistic vision is very very low, so the majority of people are corporate/government bureaucrats, impotent university drones, or modern day servants.
And this is different from history in what way? The vast majority of all people at any given time in human history have been engaged in drudgery. To think of some great human achievements: man on the moon? Takes an awful lot of nose-to-the-grindstone work to make a big-ass rocket. Pyramids? Takes slaves to drag all those rocks. And so on for millenia. The "ingenuity" and "artistic vision" gets all of the praise, but it takes more than ingenuity and vision to make things happen in this world. (Now I sound like the Marxist!)
The "cult of efficiency" of economics is interested in maximizing the amount of value that people get for their work. When it comes down to it, what's the fundamental value in the system? Human pleasure. The question of _whose_ pleasure it is, well, that's the sticky one, isn't it? It sounds like you think your pleasure is worth more than mine, or of whoever else might take your job. Thus you must be protected. Sounds like a brilliant (and morally justifiable) plan to me.
*efficient in the sense that you maximize, in whatever scale you are looking at, the values you...well, value. Other concerns (environment, "morality", etc.) might or might not be considered. Such is the nature of a maximization problem.
You can spell "pneumatic" without a hitch but you have trouble with "equivalent"???
Linux's NFS server support has gotten leaps and bounds better since about 2.4.14 or so. The "bleeding edge" NFS stuff works quite well. Is it quite up to Sun's standard? No, it's not. But it's getting close.
Of course, the perception of Linux's NFS support was probably done a fair amount of harm by Red Hat's bastardized 2.4.18 kernel that shipped with 7.3. BROKEN NFS client support out of the box with anything but Linux servers. Sent our big Sun servers into the ozone every time the load grew beyond "trivial."
If you're interested in good NFS performance, throw your Red Hat kernel away and build a clean 2.4.20, or one with the NFS patches* if you're running servers.
The patches are here.
Now, the Linux automounter (I'm talking about autofs, not amd) behaves very badly at times, but that's another story...
Clever.
But your sig refers to assonance.
There are darn few things which Sun is cost effective for anymore. Running a big DB, etc... But the word is Intel for file and print servers and smaller app servers as well. Choose a Linux box with commodity hardware and you could have that entry level blade for about $1,000. If you are worried about the reliability of commodity hardware, get a back up. You still saved half your money.
Exactly right, and this is why I think they are doing themselves serious harm by still pretending like there is profit to be made in their vertical strategy. Linux hasn't hurt Sun that badly in the computer room, but it is eating its lunch in the network room and on the desktop. Who in their right mind would shell out twice or thrice the bucks for a Sun desktop box when they can get functional equivalence or better with a Linux/Intel one?
Seems like I keep beating this horse...Sun is not going to be able to compete any more on the user end unless they join the commoditization parade. Period. They don't want to recognize this but they really need to. The no-alternative days that saw any corporation needing Unix run to one of the big vendors and forking over millions for end-to-end installations are long over. SGI has been down this path already; Sun won't be able to subsidize their desktop hardware with server revenue for much longer.
You also mention another reason I hate Solaris -- the dearth of what have become common tools and features for an OS. Yes, I know that kickback from Veritas is nice, but volume management support is about six months away from becoming a throw-in. Oh yes, it makes perfect business sense to maintain an entire separate toolchain for things like 'ls' and 'grep'. Etc.
And I really wish Sun would stop calling their shitty workstations "Blades", since that term has become accepted to mean something else entirely.
Sun's core market is not being stolen by Linux. Sun's big money comes from big servers with lots of processors. Their core market is the SunFire line (x800, and the E1xk), not the Blade line(comparable to PC).
True.
Sun's getting killed by the general reluctance of big companies to spend money on huge servers. This is in part because of general global instability (terrorism, economic downturn, imminent war, etc.), in part because there are lots of slightly-used servers floating around thanks to the dot-coms all going under, and in part because they have no idea what they are doing.
Sun is throwing money away pushing Solaris for x86 (if you can call this "pushing") and even for their workstation-class machines. Sun's throwing money away continuing to make workstation-class machines; the move away from Sun on the scientific desktop to Linux is well underway. The Blade I am stuck with is a fine machine, for a Sun, but a PC costing half as much with any recent Linux leaves it in the dust performance-wise, hardware-support-wise, and usability-wise.
Sun is in serious trouble if they don't stop wasting valuable resources on areas of business that will only become less profitable for them. I don't think they realize this, or at least they don't want to realize it.
There's a difference here: There have to be more drivers included with an x86 operating system distribution because there are so many different variations of x86 hardware, unlike with the SPARC platform where Sun Microsystems and SPARC Intl. are pretty much in control. Drivers take disc space. Granted, drivers won't fill a CD, and the Solaris OE has always lacked drivers for the newest hardware, but it's still something to be considered.
I bet SPARC is bigger and I'll tell you why -- sparcv9 is 64 bit. If Sun is still shipping 32-bit as well, then you've got the text equivalent of triple the size of the 32-bit sparc only. (I don't want to get into the relative SPARC/x86 code size war. My experience is that 32-bit sparc code is about 90% of the size of equivalent x86 code.)
I really don't think Sun is expecting to make any money on Solaris 9 anyway, so their $20 charge seems more like a disincentive for people to actually use it. Given their incredible marketing acumen and crack support team (sarcasm?), it wouldn't surprise me.
This isn't Insightful at all. Sun is not battling Microsoft in the consumer desktop market, nor does it have any reason to try. We're talking about servers here, and all MS servers* run on Intel hardware. *with the possible exception of the last 3 machines on Earth still running NT on Alpha...