At work we've been testing RAID 0 (strip) setups for performance on a number cruncher machine. Hardware raid isn't always the best.
Our first setup was 2x15kRPM U320 SCSI drives on an LSI MegaRAID controller. Apparently the 2.6 kernel driver has serious issues, because we can't get read performance over 50MB/s. This is slower than reads off a single drive on a vanilla SCSI controller.
Our second setup was the same two drives on an LSI U320 SCSI HBA. The HBA has a 'simple' stripped raid via firmware. This worked well for two drives, we got 140MB/s reads. This is about twice the single drive speed. We added a third drive to this setup and it must be swamping the firmware, because the read speeds went _down_ to 110MB/s with three drives. But when we switch to linux md software raid we got 215MB/s. This told us that we weren't yet swamping the scsi channel or PCI-X bus.
We're hoping to be able to software raid 4 15K rpm drives on a single scsi channel. This should put us around 280-300MB/s. If this works we'll try a 2 channel U320 controller with 4 drives on each channel. By offloading the raid operations from the HBA to the CPU we should be able avoid swamping the HBA processor. The only other issue to consider is weather or not we'll swap the PCI-X133 bus. If this happends well use 2 single channel controllers on seperate PCI-X133 buses. Our goal is to get over 500MB/s read on a software raid array. Really, this is an insane read speed. It is over a quarter of memory speed on these systems. We're doing this on a Dual Opteron 250.
Perhaps even bring it down safely for museum display?
Hubble was scheduled to be brought down and put in the smithsonian. In fact, the display mount is already in the air & space museum (or was a couple years ago when I visited). The problem is that the Colombia was the only shuttle decked out to down mass the Hubble. All the other orbiters are setup with an airlock and docking port for the ISS. Hubble won't fit in the cargo hold of those orbiters now.
See this is what I'm talking about. You all need to lighten up. I have no particular attachment to Linux. Over the years I've tried a variety of Unix OSs (just found a set of SCO Unixware disks the other day in fact). I use what works for me. If that's different from what other people use, fine.
I think this is why the 1080p HDTV standard isn't implimented anywhere. It just requires too much bandwidth to generate/distribute/display video at 1920x1080 @ 60fps progressive scan.
Rumor has it that Texas Instraments' next generation DLP processor will actually support the 1080p format. But I'm not holding my breath on.
If they "slightly" expanded it, though, you'd have a ~22 inch LCD with 6400x4800. Finally, a use for those high-end video cards with tons of memory!
Well, this isn't that good. But a 9 Megapixel LCD (3840x2400) would still be nice. I'd love to get two of these and run Xinerama. The problem is that each takes 2 DVI connections. So you'll need two dual DVI video cards. I'll probably have to rob a bank and wait for an Nvidia 6800 SLI system to be released.
If you have $100K of data on a single $175 drive, you, sir, are a fucking idiot. That much data (value wise) demands multiple redundencies. Hardware RAID 5 with redundent UPSs at least. A tape backup system is also highly required. And to finish it off, you ought to setup an off-site mirror.
I have a large dataset with this type of configuration. I had one of the UPSs go down today, and it released a flaw in my strategy. My rack ventilation fan was connected only to the failed ups. So don't forget to provide redundent power to your cooling system.
I find it amusing that if the missile is pointed at us we call it 'Satan'. If we point it at them, its called a 'Peacekeeper' whose role is 'Nuclear Deterence'.
Maybe they should ask Academics who are less invested in and benifited from the current patent process. Find people who are technically savy, but at least a bit removed from the patent process. Granted, many Academics recieve patents and benifit from the patent system, but I'd put money on there being fewer patent trolls amongst academics than amoung corporate executives. At least academics have obligations other than making fat sacks of cash for stock holders.
I heard a comment that since big fast disks have gotten so expensive that the I in RAID now stands for Independent. Thus alleviating the economic bottleneck.
You just need to click on the 'groups' tab in google.:) So many linux mailing lists are piped onto usenet, that just about every question asked shows up there. Essentially google groups lets you globally search nearly all linux mailing lists.
Hubble was not the first space craft to fly that size lens. When hubble was being built, Lockheed already had the equipment to test and validate the lens. As we all remember, when Hubble was put into orbit, its lens was seriously flawed and a shuttle mission had to go up and add some 'contact lens' to correct it. Now why would NASA fly an unvalidated lens when the equipment existed to validate it? Lockheed offered to do it for them, but the test equipment was in the Skunk Works, so lockheed wouldn't let any of the NASA people in without fairly hi level security clearance. None of the NASA people had the clearance and NASA didn't want to cough up the money or wait the time required to get the clearance, so they just decided not to test the lens.
I'm sure the DoD has had very high resolution stuff flying for decades. My guess is that they resolutions higher than 1cm. I went to a few technical workshops down at JPL a year or two back. There was a software contractor there who worked for the DoD on extensions to the TIFF/GeoTIFF image formats. He said they have added extentions to the TIFF format to be able to store 1PB (Peta Byte) images in a tiff file (through internally virtual images/referenced data). Multiple times he made the comment that the earth at 1cm resolution is about 1PB.
I've talked to people who worked on the Agena satelites from the 60s into the 80s. He said that though he never say the target imagery, he did see some calibration imagry in the early 70s taken over the beaches of Southern California. And yes, he could tell if they person on the beach was a man or a woman, and if a woman whether she wasy laying face up or face down. This was in the early 70s!!.
At this point I'd put money on the DoD having a constalation of satellites with far higher resolution than Hubble. On the other hand, I'm sure hubble has very different types of sesor equipment then the DoD sats.
I'm all for saving American History, but this shouldn't be NASA's job. This sort of this is the responsibility of institutions like the Smithsonian. Nasa should be spending its money on new projects. I know I'm nostalgic about the glory days of the space race, but eventually NASA will become overburdened with this sort of thing.
Slashdot Mirror
Great, now I have to go grep my kernel tree for 'dildo' just to see if there really is a kernel driver for a USB Dildo. Thanks.
At work we've been testing RAID 0 (strip) setups for performance on a number cruncher machine. Hardware raid isn't always the best.
Our first setup was 2x15kRPM U320 SCSI drives on an LSI MegaRAID controller. Apparently the 2.6 kernel driver has serious issues, because we can't get read performance over 50MB/s. This is slower than reads off a single drive on a vanilla SCSI controller.
Our second setup was the same two drives on an LSI U320 SCSI HBA. The HBA has a 'simple' stripped raid via firmware. This worked well for two drives, we got 140MB/s reads. This is about twice the single drive speed. We added a third drive to this setup and it must be swamping the firmware, because the read speeds went _down_ to 110MB/s with three drives. But when we switch to linux md software raid we got 215MB/s. This told us that we weren't yet swamping the scsi channel or PCI-X bus.
We're hoping to be able to software raid 4 15K rpm drives on a single scsi channel. This should put us around 280-300MB/s. If this works we'll try a 2 channel U320 controller with 4 drives on each channel. By offloading the raid operations from the HBA to the CPU we should be able avoid swamping the HBA processor. The only other issue to consider is weather or not we'll swap the PCI-X133 bus. If this happends well use 2 single channel controllers on seperate PCI-X133 buses. Our goal is to get over 500MB/s read on a software raid array. Really, this is an insane read speed. It is over a quarter of memory speed on these systems. We're doing this on a Dual Opteron 250.
Slashdot already covered the best method of data destruction.
Drive Slagging!
Hubble was scheduled to be brought down and put in the smithsonian. In fact, the display mount is already in the air & space museum (or was a couple years ago when I visited). The problem is that the Colombia was the only shuttle decked out to down mass the Hubble. All the other orbiters are setup with an airlock and docking port for the ISS. Hubble won't fit in the cargo hold of those orbiters now.
See this is what I'm talking about. You all need to lighten up. I have no particular attachment to Linux. Over the years I've tried a variety of Unix OSs (just found a set of SCO Unixware disks the other day in fact). I use what works for me. If that's different from what other people use, fine.
Dude, relax. It was a joke. The only thing funnier than jokes about BSD dying is the insecurity of BSD people when you joke about it dying.
Oh come on people. We all know that BSD is dead. This article is a complete lie. It just has to be!
The JungleBoy
To quote someone whom I can't remember:
"The nipple is the ONLY intuitive interface. After that, it's all learned"
I think this is why the 1080p HDTV standard isn't implimented anywhere. It just requires too much bandwidth to generate/distribute/display video at 1920x1080 @ 60fps progressive scan.
Rumor has it that Texas Instraments' next generation DLP processor will actually support the 1080p format. But I'm not holding my breath on.
-JungleBoy
Favorite quote has to be from one of the PSAs:
"Dude, don't taunt him. He has a graph."
Come on people, get to work! I'm pulling down the episode 39 divx file at 500KB/s (yes, big B). What a bunch of slackers. Get clickin'
Well, this isn't that good. But a 9 Megapixel LCD (3840x2400) would still be nice. I'd love to get two of these and run Xinerama. The problem is that each takes 2 DVI connections. So you'll need two dual DVI video cards. I'll probably have to rob a bank and wait for an Nvidia 6800 SLI system to be released.
-JungleBoyIf you have $100K of data on a single $175 drive, you, sir, are a fucking idiot. That much data (value wise) demands multiple redundencies. Hardware RAID 5 with redundent UPSs at least. A tape backup system is also highly required. And to finish it off, you ought to setup an off-site mirror.
I have a large dataset with this type of configuration. I had one of the UPSs go down today, and it released a flaw in my strategy. My rack ventilation fan was connected only to the failed ups. So don't forget to provide redundent power to your cooling system.
Now would be a great time for me to direct you all towards this illustrated commentary: Google = Money
-JungleBoy
Or if you take the heatsink off and press someones face against it. It could be some new torture device.
I find it amusing that if the missile is pointed at us we call it 'Satan'. If we point it at them, its called a 'Peacekeeper' whose role is 'Nuclear Deterence'.
> Who else should they ask?
Maybe they should ask Academics who are less invested in and benifited from the current patent process. Find people who are technically savy, but at least a bit removed from the patent process. Granted, many Academics recieve patents and benifit from the patent system, but I'd put money on there being fewer patent trolls amongst academics than amoung corporate executives. At least academics have obligations other than making fat sacks of cash for stock holders.
I find that it doesn't matter what type of beans you use, as long as you make it with Water Joe and grind up a hand full of NoDoze with the beans.
Thats the sort of thing that gets me going... And probably leads to situations like my signature.
-JungleBoy
I don't think I could say that for myself because 10 x 0 is still zero.
I would have that simply say that I am productive when I don't read /. at work. When I read /. at work, my productivity isn't mostly dead, its all dead.
I heard a comment that since big fast disks have gotten so expensive that the I in RAID now stands for Independent. Thus alleviating the economic bottleneck.
You just need to click on the 'groups' tab in google. :) So many linux mailing lists are piped onto usenet, that just about every question asked shows up there. Essentially google groups lets you globally search nearly all linux mailing lists.
I wonder how much Fossil Fuel is needed to produce the Ethanol? I seriously doubt this is a truly Fossil Fuel Free[tm] method of making Hydrogen fuel.
Hubble was not the first space craft to fly that size lens. When hubble was being built, Lockheed already had the equipment to test and validate the lens. As we all remember, when Hubble was put into orbit, its lens was seriously flawed and a shuttle mission had to go up and add some 'contact lens' to correct it. Now why would NASA fly an unvalidated lens when the equipment existed to validate it? Lockheed offered to do it for them, but the test equipment was in the Skunk Works, so lockheed wouldn't let any of the NASA people in without fairly hi level security clearance. None of the NASA people had the clearance and NASA didn't want to cough up the money or wait the time required to get the clearance, so they just decided not to test the lens.
I'm sure the DoD has had very high resolution stuff flying for decades. My guess is that they resolutions higher than 1cm. I went to a few technical workshops down at JPL a year or two back. There was a software contractor there who worked for the DoD on extensions to the TIFF/GeoTIFF image formats. He said they have added extentions to the TIFF format to be able to store 1PB (Peta Byte) images in a tiff file (through internally virtual images/referenced data). Multiple times he made the comment that the earth at 1cm resolution is about 1PB.
I've talked to people who worked on the Agena satelites from the 60s into the 80s. He said that though he never say the target imagery, he did see some calibration imagry in the early 70s taken over the beaches of Southern California. And yes, he could tell if they person on the beach was a man or a woman, and if a woman whether she wasy laying face up or face down. This was in the early 70s!!.
At this point I'd put money on the DoD having a constalation of satellites with far higher resolution than Hubble. On the other hand, I'm sure hubble has very different types of sesor equipment then the DoD sats.
I'm all for saving American History, but this shouldn't be NASA's job. This sort of this is the responsibility of institutions like the Smithsonian. Nasa should be spending its money on new projects. I know I'm nostalgic about the glory days of the space race, but eventually NASA will become overburdened with this sort of thing.
"Never argue with an idiot, he'll just lower you to his level and beat you with experience."
-JungleBoy