That 99th percentile is based on barely enough drives for it to be rated (just over 60) so it doesn't mean much, most of the drives have hundreds of units in the database. Besides I think the database is in many ways flawed as most people who list their drives will do so because they have had a failed drive. The best way from my perspective is to look at what companies with hundreds or thousands of drives are doing. Rackspace switched out all of their IBM IDE's for Maxtors, google uses Maxtor's, and the recent IDE backup unit featured here used Maxtors. But maybe I'm just biased because as an OEM I had tons of drives from almost every manufacturer die on me except for Maxtors.
Yep that's the difference between CS and the real world. The x86 ISA may not be the most elegant or clean but it is kicking the snot out of everything except maybe Power. Sure it can be seen as kludge on kludge but yet no one seems to be able to come out with something that beats it for perfomance without costing many times more.
It's probably a bit of both, it is probably very similar to FX!32 for the Dec Alpha version of NT4. What this did was emulated many x86 functions, but if something was getting called a lot it was dynamically recompiled to native Alpha code. Worked pretty well overall.
Umm if you simply ASK most employers will not have a problem with releasing the work, they got the tool they asked for, so what if someone else can use the same tool. Unless you are in the business of selling software it often makes sense to release your tools, they may be maintained for free =0 Yes there are PHB's or companies with specific reasons for not allowing tools to be released OS, but I would bet they are in the minority if you simply took that first step =)
Another common one is that they have a tool that gets the job done a hoarding it does them no good, so they release it for the common good. Of course most of them do it because they found value in another open source application they used and figure "hey I got something for nothing why not release this if there is a chance it will help someone." A good example of this from my personal experience is CEPS or Cisco Enterprise Print System, while there is arguably not a lot of new code there (it is based around a number of open source tools), the completed package is definitly worth more than the sum of its parts. The author was very happy at finding open and free solutions he could use to get his job done, and in return released the best print system in existance back to the world for all to use. The author gained something from the open source movement (all the free tools that allowed him to make a super low cost print system that beats every commercial system out there) and the community gained something (this great tool). Everyone wins and it costs him almost no additional time or expense to release his work as open source. To check out the project go to the CEPS page at sourceforge.net
I really really doubt you are likely to wear out a card storing media (exactly how often do you change out your music? Even every day would put you at 1,000+ years =) Using is for swap on the other hand....
Because they have limited resources with which to teach a future generation of doctors which will be doing more for this society then 90+% of people? This is not s socialist's viewpoint but a capitalists viewpoint, it is ineffecient to waste resources when they could be applied to productive outputs. The cost of a medical education is just a downpayment on a future ROI.
No, because anyone who puts dogma before logic is not going to be very good at using deductive reasoning in a critical situation if their dogma gets in the way (say a mother is dying due to rejection problems with a fetus and the only way to save her is removal of the fetus and it is too young to survive outside the mother, what does the person do? logic dictates that you remove the fetus because it will die anyways if the mother dies but I have been told stories by a doctor friend of mine of colleagues refusing based on personal beliefs that reject abortion in any case and saying things like "if the lord want her to survive she will and it will save the unborn life"). I don't think that religious people in general would make bad doctors, quite the opposite, but those who are so embroiled in their faith as to reject logic are.
Why use Adaptec's cards, they are much lower performing and much less feature rich then the 3Ware cards. Plus you are limited to 4 ports so you are always losing 25% capacity to parity, or 50% if you want hotspare.
3Ware controllers support hotswap and hotspare, so on failure the data is recreated to the spare so need to rush in replacing the drive (in other words get around to it whenever you have time that week) and then you just unplug the drive and plop in a new one. Plus they have lost a total of 3 drives in the first year, I had to change out tapes every 2 weeks, this is a lot less work =)
Because I would be willing to bet that Quantum will charge a hell of a lot more per TB then this system cost (69TB for less than half a million is a bargain).
I don't think you'd be putting in 100 lines, more like whatever a standard bundle contains and using gigabit or 10 gigabit over it. The advantage of this is that you can use it for other things =) 50K doesn't buy you much of a tape library. Look into the cost of a StorageTek that can handle more than 4 drives and more than 100 tapes, they cost a LOT more than $50K.
No the current high end tape solution stores 500GB natively and over a TB compressed (Sony Super AIT). With SDLT at 160GB native 320GB compressed. Transfer rates for both are around 30MB/s native 60MB/s compressed. Your data sounds like old DLT IV numbers (admittadly those are probably the majority in use). This is for cartridge style solutions, reel to reel are even larger capacity.
Did you read the article, 0.53% failure rate for their hdd's and 0% failure at the node and system level. RAID5 + hotspare is pretty damn good at protecting data assuming you have reliable clean power. Look into what it costs to get a storagetek that can handle 100's of TB's and then compare to the less than half million spent on this system. Then add in admin time, training, and the additional time backups and restores take. I think that this type of backup system is really going to take off over the next couple of years, every backup and storage vendor has something similar out or in the works. Of course those with the most demanding of applications will use this in a multilayered aproach where they backup the live data to the ATA cluster for speed and then stream it to tape during the day for offsite backup, or they will have another unit somewhere else and will use large networking pipes to mirror the data there.
Good point, he does talk about 100 nodes, why not have them on seperate ends of campus or even across town. Using longhaul fibre adapters they could go up 16 miles I believe without a repeater. So just devide the nodes into two groups and mirror the data to both sites, still be cheaper than tape. Sure it wouldn't work for a multinational corporation (for instance the telephone and transmitters in NY were often mirrored by being in each of the twin towers, this is now seen as being "not a good idea") but anything that takes out both ends of campus or two ends of town is probably so big that the universities last concerns will be the backup data.
Well as the article states this implementation isn't really for offsite There's one aspect in which Dr. Koch's backup system can't keep up with tape solutions: storing the backup medium in another location after the backup has been completed. but it could be done pretty easily. Non-operating shock capacity on the D540X is 300G's for 2ms which is pretty darn good (plastic tape housings might shatter under a similar load). I also like the ultra low failure rate.5% (hmm, this and the data from storage review shows that the D540X and D740X line seem to be some of the most reliable out there...) I know our DAT failure rate was in the same ballpark.
clean your vinyl and use decent needles and you won't have much problem. My dad has 35 year old LP's that still sound damn good today because they were well cared for. Actually I want to get a laser record player so I can have the best of both worlds, the nice analog form of LP's with the replayability of cd's. That and real DJ's will never switch to cd, it just isn't anywhere near the same thing. Mixing with vinyl is a tactile experience that none of the cd scratch decks comes close to replicating (I know I've tried most of the "pro" ones.)
Yes it is a competition, with a decent quality soundcard and heaphone amp I can't tell the difference between 44.1Khz 16bit wav's and ~200Kbps VBR mp3's from Lame using my Sennheisers for ~95+% of my music collection. I did true A-B-C blind tests between LAME, wav, and OOG and I could only pick out OOG on most samples.
Get large systems fron Dell, HP, IBM, or Unisys and you get large memory support, up to 64GB for most of them on the largest x86 servers. The low end 1 and 2 way chipsets don't support large amounts of memory because PAE support takes a lot of silicon that just isn't needed by 99.9999% of the people who buy such systems. That and you'd need more slots as the largest commonly available ram modules are 1GB, though 4GB modules are seen a little more often these days.
Ummm, large X86 systems from Dell, HP, Unisys and others already support the full 64GB of ram that the architecture allows. Sure not everyone (or even many) who uses Linux needs large PAE support but it would be nice for those who do. Of course most of the people I know who wanted to use large amounts of ram on x86 are waiting for Opteron because generally if you need that much ram you need it all for one or two processes and that doesn't work with PAE. btw if you think no one wants access to large chunks of ram on x86 then you haven't dealt with people with large databases or who route large ASIC's/cpus. It costs shedloads to buy Sun's with lots of ram, an Opteron or Xeon with the same amount of ram will probably cost about 1/3rd as much, which when you are talking about dozens of systems adds up to a lot of money. (My back of the envelope calculation was that for our ASIC group it would save somewhere in the neighborhood of about a quarter million a year =)
Actually there was a case where a gentleman in the UK was cited for tax evasion for using homebrew biodiesel because his fuel was not taxed and hence he was not doing his part to maintain the roads.
clearcase is great and horrible at the same time. It's great because it does its job well and never corrupts data. It's horrible because its expensive, requires HUGE expensive machines to run on, and just isn't all that pretty in some areas. We "upgraded" from cvs to clearcase and the server went from a spare P2 class rack server to some monsterous Sun machine, the developers were not ammused as they would have much rather kept the old box and spent the money on new workstations.
Slashdot Mirror
That 99th percentile is based on barely enough drives for it to be rated (just over 60) so it doesn't mean much, most of the drives have hundreds of units in the database. Besides I think the database is in many ways flawed as most people who list their drives will do so because they have had a failed drive. The best way from my perspective is to look at what companies with hundreds or thousands of drives are doing. Rackspace switched out all of their IBM IDE's for Maxtors, google uses Maxtor's, and the recent IDE backup unit featured here used Maxtors. But maybe I'm just biased because as an OEM I had tons of drives from almost every manufacturer die on me except for Maxtors.
Yep that's the difference between CS and the real world. The x86 ISA may not be the most elegant or clean but it is kicking the snot out of everything except maybe Power. Sure it can be seen as kludge on kludge but yet no one seems to be able to come out with something that beats it for perfomance without costing many times more.
It's probably a bit of both, it is probably very similar to FX!32 for the Dec Alpha version of NT4. What this did was emulated many x86 functions, but if something was getting called a lot it was dynamically recompiled to native Alpha code. Worked pretty well overall.
Umm if you simply ASK most employers will not have a problem with releasing the work, they got the tool they asked for, so what if someone else can use the same tool. Unless you are in the business of selling software it often makes sense to release your tools, they may be maintained for free =0 Yes there are PHB's or companies with specific reasons for not allowing tools to be released OS, but I would bet they are in the minority if you simply took that first step =)
Another common one is that they have a tool that gets the job done a hoarding it does them no good, so they release it for the common good. Of course most of them do it because they found value in another open source application they used and figure "hey I got something for nothing why not release this if there is a chance it will help someone." A good example of this from my personal experience is CEPS or Cisco Enterprise Print System, while there is arguably not a lot of new code there (it is based around a number of open source tools), the completed package is definitly worth more than the sum of its parts. The author was very happy at finding open and free solutions he could use to get his job done, and in return released the best print system in existance back to the world for all to use. The author gained something from the open source movement (all the free tools that allowed him to make a super low cost print system that beats every commercial system out there) and the community gained something (this great tool). Everyone wins and it costs him almost no additional time or expense to release his work as open source. To check out the project go to the CEPS page at sourceforge.net
I really really doubt you are likely to wear out a card storing media (exactly how often do you change out your music? Even every day would put you at 1,000+ years =) Using is for swap on the other hand....
Do not use microdrives, just get a 1GB CF card, much much lower power consumption, faster transfer, and at this point cheaper most places.
The FBI and the ATF??
Because they have limited resources with which to teach a future generation of doctors which will be doing more for this society then 90+% of people? This is not s socialist's viewpoint but a capitalists viewpoint, it is ineffecient to waste resources when they could be applied to productive outputs. The cost of a medical education is just a downpayment on a future ROI.
No, because anyone who puts dogma before logic is not going to be very good at using deductive reasoning in a critical situation if their dogma gets in the way (say a mother is dying due to rejection problems with a fetus and the only way to save her is removal of the fetus and it is too young to survive outside the mother, what does the person do? logic dictates that you remove the fetus because it will die anyways if the mother dies but I have been told stories by a doctor friend of mine of colleagues refusing based on personal beliefs that reject abortion in any case and saying things like "if the lord want her to survive she will and it will save the unborn life"). I don't think that religious people in general would make bad doctors, quite the opposite, but those who are so embroiled in their faith as to reject logic are.
Why not ATX, almost all dual processor servers outside of the big OEM's are based on ATX boards.
Why use Adaptec's cards, they are much lower performing and much less feature rich then the 3Ware cards. Plus you are limited to 4 ports so you are always losing 25% capacity to parity, or 50% if you want hotspare.
3Ware controllers support hotswap and hotspare, so on failure the data is recreated to the spare so need to rush in replacing the drive (in other words get around to it whenever you have time that week) and then you just unplug the drive and plop in a new one. Plus they have lost a total of 3 drives in the first year, I had to change out tapes every 2 weeks, this is a lot less work =)
Because I would be willing to bet that Quantum will charge a hell of a lot more per TB then this system cost (69TB for less than half a million is a bargain).
I don't think you'd be putting in 100 lines, more like whatever a standard bundle contains and using gigabit or 10 gigabit over it. The advantage of this is that you can use it for other things =) 50K doesn't buy you much of a tape library. Look into the cost of a StorageTek that can handle more than 4 drives and more than 100 tapes, they cost a LOT more than $50K.
No the current high end tape solution stores 500GB natively and over a TB compressed (Sony Super AIT). With SDLT at 160GB native 320GB compressed. Transfer rates for both are around 30MB/s native 60MB/s compressed. Your data sounds like old DLT IV numbers (admittadly those are probably the majority in use). This is for cartridge style solutions, reel to reel are even larger capacity.
Did you read the article, 0.53% failure rate for their hdd's and 0% failure at the node and system level. RAID5 + hotspare is pretty damn good at protecting data assuming you have reliable clean power. Look into what it costs to get a storagetek that can handle 100's of TB's and then compare to the less than half million spent on this system. Then add in admin time, training, and the additional time backups and restores take. I think that this type of backup system is really going to take off over the next couple of years, every backup and storage vendor has something similar out or in the works. Of course those with the most demanding of applications will use this in a multilayered aproach where they backup the live data to the ATA cluster for speed and then stream it to tape during the day for offsite backup, or they will have another unit somewhere else and will use large networking pipes to mirror the data there.
Good point, he does talk about 100 nodes, why not have them on seperate ends of campus or even across town. Using longhaul fibre adapters they could go up 16 miles I believe without a repeater. So just devide the nodes into two groups and mirror the data to both sites, still be cheaper than tape. Sure it wouldn't work for a multinational corporation (for instance the telephone and transmitters in NY were often mirrored by being in each of the twin towers, this is now seen as being "not a good idea") but anything that takes out both ends of campus or two ends of town is probably so big that the universities last concerns will be the backup data.
Well as the article states this implementation isn't really for offsite There's one aspect in which Dr. Koch's backup system can't keep up with tape solutions: storing the backup medium in another location after the backup has been completed. but it could be done pretty easily. Non-operating shock capacity on the D540X is 300G's for 2ms which is pretty darn good (plastic tape housings might shatter under a similar load). I also like the ultra low failure rate .5% (hmm, this and the data from storage review shows that the D540X and D740X line seem to be some of the most reliable out there...) I know our DAT failure rate was in the same ballpark.
clean your vinyl and use decent needles and you won't have much problem. My dad has 35 year old LP's that still sound damn good today because they were well cared for. Actually I want to get a laser record player so I can have the best of both worlds, the nice analog form of LP's with the replayability of cd's. That and real DJ's will never switch to cd, it just isn't anywhere near the same thing. Mixing with vinyl is a tactile experience that none of the cd scratch decks comes close to replicating (I know I've tried most of the "pro" ones.)
Yes it is a competition, with a decent quality soundcard and heaphone amp I can't tell the difference between 44.1Khz 16bit wav's and ~200Kbps VBR mp3's from Lame using my Sennheisers for ~95+% of my music collection. I did true A-B-C blind tests between LAME, wav, and OOG and I could only pick out OOG on most samples.
Get large systems fron Dell, HP, IBM, or Unisys and you get large memory support, up to 64GB for most of them on the largest x86 servers. The low end 1 and 2 way chipsets don't support large amounts of memory because PAE support takes a lot of silicon that just isn't needed by 99.9999% of the people who buy such systems. That and you'd need more slots as the largest commonly available ram modules are 1GB, though 4GB modules are seen a little more often these days.
Ummm, large X86 systems from Dell, HP, Unisys and others already support the full 64GB of ram that the architecture allows. Sure not everyone (or even many) who uses Linux needs large PAE support but it would be nice for those who do. Of course most of the people I know who wanted to use large amounts of ram on x86 are waiting for Opteron because generally if you need that much ram you need it all for one or two processes and that doesn't work with PAE. btw if you think no one wants access to large chunks of ram on x86 then you haven't dealt with people with large databases or who route large ASIC's/cpus. It costs shedloads to buy Sun's with lots of ram, an Opteron or Xeon with the same amount of ram will probably cost about 1/3rd as much, which when you are talking about dozens of systems adds up to a lot of money. (My back of the envelope calculation was that for our ASIC group it would save somewhere in the neighborhood of about a quarter million a year =)
Actually there was a case where a gentleman in the UK was cited for tax evasion for using homebrew biodiesel because his fuel was not taxed and hence he was not doing his part to maintain the roads.
clearcase is great and horrible at the same time. It's great because it does its job well and never corrupts data. It's horrible because its expensive, requires HUGE expensive machines to run on, and just isn't all that pretty in some areas. We "upgraded" from cvs to clearcase and the server went from a spare P2 class rack server to some monsterous Sun machine, the developers were not ammused as they would have much rather kept the old box and spent the money on new workstations.