RE: Hard disk technology (was: Disk based backup)
I suspect this new drive claiming 140MB/sec sustained rates is running 2 heads in parallel since this number is about double what you see in normal drives. They could have sped up the rotational speed and/or upped the bit density to get this performance as well. -Original Message- From: Ben Scott [mailto:mailvor...@gmail.com] Sent: Monday, September 21, 2009 12:13 PM To: NT System Admin Issues Subject: Re: Hard disk technology (was: Disk based backup) On Mon, Sep 21, 2009 at 11:09 AM, Alverson, Tom (Xetron) tom.alver...@ngc.com wrote: I'm wondering why they have not done this yet as well. Using more that 2 heads in parallel would at some point be enough to saturate existing SATA interfaces. Unless they *are* already doing it, and the current speeds of hard drives reflect that. Do you have any reason to believe they are not already doing so? -- Ben ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~ ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~
RE: Hard disk technology (was: Disk based backup)
Correct. Until they improve the raw speed of the drives, the faster interface helps mainly to speed up reading and writing to the drives onboard ram cache, which is a good thing but doesn't help applications where you need sustained high speed reading and writing (video, raw data collection etc). -Original Message- From: Ben Scott [mailto:mailvor...@gmail.com] Sent: Monday, September 21, 2009 2:18 PM To: NT System Admin Issues Subject: Re: Hard disk technology (was: Disk based backup) On Mon, Sep 21, 2009 at 2:08 PM, Sam Cayze sam.ca...@rollouts.com wrote: It promises a sustained transfer rate of just 140MBps ... I presume they mean 140 Mbyte/sec, which is 1140 Mbit/sec (ignoring overhead). In other words, it can't even saturate first-generation SATA (1500 Mbit/sec), let alone second-generation (3000 Mbit/sec), let twice alone the new 6000 Mbit/sec some people are so excited about. -- Ben ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~ ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~
RE: Hard disk technology (was: Disk based backup)
I'm wondering why they have not done this yet as well. Using more that 2 heads in parallel would at some point be enough to saturate existing SATA interfaces. There is no way they could do 8 heads in parallel and have enough bandwidth (even with 3.0 Gbit/sec SATA) to prevent saturating the SATA bus. Flash hard drives have been doing this since the first ones came out several years ago (running multiple sub-systems in parallel) to overcome the slower speeds of the Flash chips. I don't know if the modern flash hard drives need to do this any more (the flash chips might be fast enough now). The first Flash hard drives ran about $30,000 for a 30GB drive, and they were in the 3.5 in form factor to fit all the chips. Tom -Original Message- From: Ben Scott [mailto:mailvor...@gmail.com] Sent: Thursday, September 17, 2009 8:37 PM To: NT System Admin Issues Subject: Hard disk technology (was: Disk based backup) On Thu, Sep 17, 2009 at 4:10 PM, Alverson, Tom (Xetron) tom.alver...@ngc.com wrote: Modern hard drives can sustain, what, maybe 0.4 to 0.6 Gbit/sec? Even the 3 Gbit/sec we have now is much higher than that. How is moving to 6 Gbit/sec going to help? :) All they need to do is upgrade the (on-board) controllers to operate all the heads in parallel. For a 4 platter drive (8 heads) they could get an immediate 8X improvement in real read and write speeds. But that's got nothing to do with any particular revision of the SATA standard, right? You can interleave data across heads whether it's SATA, SCSI, or PATA. Even PATA hard drives have been presenting synthetic C/H/S geometry for something like a decade now, so it's not like the controllers don't already have to do the job. And given that *current* hard drives can't even keep up with SATA at 1.5 Gbit/sec, I would then ask: Why haven't they done this already? Or have they already, and the speeds we see today are *with* interleaving? (I'll ignore for the time being that most consumer hard disks these days only have one or two heads anyway.) -- Ben ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~ ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~
RE: Hard disk technology (was: Disk based backup)
It may have been done already (2 heads running in parallel). I have a Seagate 1.5TB drive that is surprisingly fast (115MB/sec sustained) that may be doing this, or they may just have a very high bit density. The 750GB drives we have run about 70MB/sec with the same test. Tom -Original Message- From: Ken Schaefer [mailto:k...@adopenstatic.com] Sent: Thursday, September 17, 2009 11:01 PM To: NT System Admin Issues Subject: RE: Hard disk technology (was: Disk based backup) I have to agree with Ben here. If it were easy to do, it would have been done already. I suspect the improved bus speeds will help devices that aren't current spinning disks (maybe flash based drives), or where we are able to present an array of disks at the end of the bus (e.g. external direct attached storage connected via eSATA) Cheers Ken ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~
Re: Hard disk technology (was: Disk based backup)
I'm very curious as to when did harddrive manufacturers swich to having independently operating heads? Is this a new development that I missed? For as long as I can remember, the heads (regardless of how many there are) are operated by a single head motor. So while it can interleave data across multiple platters, all the heads would still be at the same location on the platters. On Mon, Sep 21, 2009 at 11:09 AM, Alverson, Tom (Xetron) tom.alver...@ngc.com wrote: I'm wondering why they have not done this yet as well. Using more that 2 heads in parallel would at some point be enough to saturate existing SATA interfaces. There is no way they could do 8 heads in parallel and have enough bandwidth (even with 3.0 Gbit/sec SATA) to prevent saturating the SATA bus. Flash hard drives have been doing this since the first ones came out several years ago (running multiple sub-systems in parallel) to overcome the slower speeds of the Flash chips. I don't know if the modern flash hard drives need to do this any more (the flash chips might be fast enough now). The first Flash hard drives ran about $30,000 for a 30GB drive, and they were in the 3.5 in form factor to fit all the chips. Tom -Original Message- From: Ben Scott [mailto:mailvor...@gmail.com] Sent: Thursday, September 17, 2009 8:37 PM To: NT System Admin Issues Subject: Hard disk technology (was: Disk based backup) On Thu, Sep 17, 2009 at 4:10 PM, Alverson, Tom (Xetron) tom.alver...@ngc.com wrote: Modern hard drives can sustain, what, maybe 0.4 to 0.6 Gbit/sec? Even the 3 Gbit/sec we have now is much higher than that. How is moving to 6 Gbit/sec going to help? :) All they need to do is upgrade the (on-board) controllers to operate all the heads in parallel. For a 4 platter drive (8 heads) they could get an immediate 8X improvement in real read and write speeds. But that's got nothing to do with any particular revision of the SATA standard, right? You can interleave data across heads whether it's SATA, SCSI, or PATA. Even PATA hard drives have been presenting synthetic C/H/S geometry for something like a decade now, so it's not like the controllers don't already have to do the job. And given that *current* hard drives can't even keep up with SATA at 1.5 Gbit/sec, I would then ask: Why haven't they done this already? Or have they already, and the speeds we see today are *with* interleaving? (I'll ignore for the time being that most consumer hard disks these days only have one or two heads anyway.) -- Ben ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~ ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~ ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~
RE: Hard disk technology (was: Disk based backup)
But this info is not in the specs of thee HD isn't it? So how do you know that is actually that fast before buying it? Miguel --- El lun, 21/9/09, Alverson, Tom (Xetron) tom.alver...@ngc.com escribió: De: Alverson, Tom (Xetron) tom.alver...@ngc.com Asunto: RE: Hard disk technology (was: Disk based backup) Para: NT System Admin Issues ntsysadmin@lyris.sunbelt-software.com Fecha: lunes, 21 septiembre, 2009 11:13 It may have been done already (2 heads running in parallel). I have a Seagate 1.5TB drive that is surprisingly fast (115MB/sec sustained) that may be doing this, or they may just have a very high bit density. The 750GB drives we have run about 70MB/sec with the same test. Tom -Original Message- From: Ken Schaefer [mailto:k...@adopenstatic.com] Sent: Thursday, September 17, 2009 11:01 PM To: NT System Admin Issues Subject: RE: Hard disk technology (was: Disk based backup) I have to agree with Ben here. If it were easy to do, it would have been done already. I suspect the improved bus speeds will help devices that aren't current spinning disks (maybe flash based drives), or where we are able to present an array of disks at the end of the bus (e.g. external direct attached storage connected via eSATA) Cheers Ken ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~ ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~
RE: Hard disk technology (was: Disk based backup)
The heads are all locked together. There is no need to have them move independently. From: Scott Kaufman [mailto:bskauf...@gmail.com] Sent: Monday, September 21, 2009 11:17 AM To: NT System Admin Issues Subject: Re: Hard disk technology (was: Disk based backup) I'm very curious as to when did harddrive manufacturers swich to having independently operating heads? Is this a new development that I missed? For as long as I can remember, the heads (regardless of how many there are) are operated by a single head motor. So while it can interleave data across multiple platters, all the heads would still be at the same location on the platters. On Mon, Sep 21, 2009 at 11:09 AM, Alverson, Tom (Xetron) tom.alver...@ngc.com wrote: I'm wondering why they have not done this yet as well. Using more that 2 heads in parallel would at some point be enough to saturate existing SATA interfaces. There is no way they could do 8 heads in parallel and have enough bandwidth (even with 3.0 Gbit/sec SATA) to prevent saturating the SATA bus. Flash hard drives have been doing this since the first ones came out several years ago (running multiple sub-systems in parallel) to overcome the slower speeds of the Flash chips. I don't know if the modern flash hard drives need to do this any more (the flash chips might be fast enough now). The first Flash hard drives ran about $30,000 for a 30GB drive, and they were in the 3.5 in form factor to fit all the chips. Tom -Original Message- From: Ben Scott [mailto:mailvor...@gmail.com] Sent: Thursday, September 17, 2009 8:37 PM To: NT System Admin Issues Subject: Hard disk technology (was: Disk based backup) On Thu, Sep 17, 2009 at 4:10 PM, Alverson, Tom (Xetron) tom.alver...@ngc.com wrote: Modern hard drives can sustain, what, maybe 0.4 to 0.6 Gbit/sec? Even the 3 Gbit/sec we have now is much higher than that. How is moving to 6 Gbit/sec going to help? :) All they need to do is upgrade the (on-board) controllers to operate all the heads in parallel. For a 4 platter drive (8 heads) they could get an immediate 8X improvement in real read and write speeds. But that's got nothing to do with any particular revision of the SATA standard, right? You can interleave data across heads whether it's SATA, SCSI, or PATA. Even PATA hard drives have been presenting synthetic C/H/S geometry for something like a decade now, so it's not like the controllers don't already have to do the job. And given that *current* hard drives can't even keep up with SATA at 1.5 Gbit/sec, I would then ask: Why haven't they done this already? Or have they already, and the speeds we see today are *with* interleaving? (I'll ignore for the time being that most consumer hard disks these days only have one or two heads anyway.) -- Ben ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~ ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~ ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~
RE: Hard disk technology (was: Disk based backup)
There is if you want to read from different portions of different platters... And write performance is nowhere near the speed of read performance at the moment. Cheers Ken From: Alverson, Tom (Xetron) [mailto:tom.alver...@ngc.com] Sent: Tuesday, 22 September 2009 1:54 AM To: NT System Admin Issues Subject: RE: Hard disk technology (was: Disk based backup) The heads are all locked togethe There is no need to have them move independently. From: Scott Kaufman [mailto:bskauf...@gmail.com] Sent: Monday, September 21, 2009 11:17 AM To: NT System Admin Issues Subject: Re: Hard disk technology (was: Disk based backup) I'm very curious as to when did harddrive manufacturers swich to having independently operating heads? Is this a new development that I missed? For as long as I can remember, the heads (regardless of how many there are) are operated by a single head motor. So while it can interleave data across multiple platters, all the heads would still be at the same location on the platters. On Mon, Sep 21, 2009 at 11:09 AM, Alverson, Tom (Xetron) tom.alver...@ngc.commailto:tom.alver...@ngc.com wrote: I'm wondering why they have not done this yet as well. Using more that 2 heads in parallel would at some point be enough to saturate existing SATA interfaces. There is no way they could do 8 heads in parallel and have enough bandwidth (even with 3.0 Gbit/sec SATA) to prevent saturating the SATA bus. Flash hard drives have been doing this since the first ones came out several years ago (running multiple sub-systems in parallel) to overcome the slower speeds of the Flash chips. I don't know if the modern flash hard drives need to do this any more (the flash chips might be fast enough now). The first Flash hard drives ran about $30,000 for a 30GB drive, and they were in the 3.5 in form factor to fit all the chips. Tom -Original Message- From: Ben Scott [mailto:mailvor...@gmail.commailto:mailvor...@gmail.com] Sent: Thursday, September 17, 2009 8:37 PM To: NT System Admin Issues Subject: Hard disk technology (was: Disk based backup) On Thu, Sep 17, 2009 at 4:10 PM, Alverson, Tom (Xetron) tom.alver...@ngc.commailto:tom.alver...@ngc.com wrote: Modern hard drives can sustain, what, maybe 0.4 to 0.6 Gbit/sec? Even the 3 Gbit/sec we have now is much higher than that. How is moving to 6 Gbit/sec going to help? :) All they need to do is upgrade the (on-board) controllers to operate all the heads in parallel. For a 4 platter drive (8 heads) they could get an immediate 8X improvement in real read and write speeds. But that's got nothing to do with any particular revision of the SATA standard, right? You can interleave data across heads whether it's SATA, SCSI, or PATA. Even PATA hard drives have been presenting synthetic C/H/S geometry for something like a decade now, so it's not like the controllers don't already have to do the job. And given that *current* hard drives can't even keep up with SATA at 1.5 Gbit/sec, I would then ask: Why haven't they done this already? Or have they already, and the speeds we see today are *with* interleaving? (I'll ignore for the time being that most consumer hard disks these days only have one or two heads anyway.) ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~
Re: Hard disk technology (was: Disk based backup)
On Mon, Sep 21, 2009 at 11:17 AM, Scott Kaufman bskauf...@gmail.com wrote: So while it can interleave data across multiple platters, all the heads would still be at the same location on the platters. In a sequential read or write operation, by interleaving data across the platters (actually, sides of platters), you should be able to increase throughput. If you don't interleave this way, then you read all sectors along a single track, then seek to the next track. Once you've covered all tracks on a side, you move to the next side (head). With this interleaving, you read all sectors on all sides at once. Then you read the next set of sectors, all at once. When you reach the end of the track, you seek to the next track, but keep reading all heads at once. You essentially multiply sequential throughput by the number of sides. Same for writing. If you're doing random I/O instead, you don't get any benefit from this, but it doesn't hurt anything, either. You already have to seek -- move the arm and wait for the sector to rotate around -- anyway. Just only activate the head you need. This would require the controller have logic to drive and buffer multiple heads at once (without this kind of interleaving, you could switch a single set of logic across all heads), but beyond that, it should be trivial. As I said, it's entirely possible, even likely, they do this already. -- Ben ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~
Re: Hard disk technology (was: Disk based backup)
On Mon, Sep 21, 2009 at 11:09 AM, Alverson, Tom (Xetron) tom.alver...@ngc.com wrote: I'm wondering why they have not done this yet as well. Using more that 2 heads in parallel would at some point be enough to saturate existing SATA interfaces. Unless they *are* already doing it, and the current speeds of hard drives reflect that. Do you have any reason to believe they are not already doing so? -- Ben ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~
Re: Hard disk technology (was: Disk based backup)
On Mon, Sep 21, 2009 at 2:08 PM, Sam Cayze sam.ca...@rollouts.com wrote: It promises a sustained transfer rate of just 140MBps ... I presume they mean 140 Mbyte/sec, which is 1140 Mbit/sec (ignoring overhead). In other words, it can't even saturate first-generation SATA (1500 Mbit/sec), let alone second-generation (3000 Mbit/sec), let twice alone the new 6000 Mbit/sec some people are so excited about. -- Ben ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~
RE: Hard disk technology (was: Disk based backup)
Related: http://www.engadget.com/2009/09/21/seagate-2tb-barracuda-xt-worlds-first -sata-6gbps-hard-drive/ Haven't been following the thread, but just saw this on Engadget. Ready for this speed freaks? Seagate just announced the world's first 2TB disk with full support for the third generation SATA interface pushing data at 6Gbps -- double the rate of previous controllers. The 3.5-inch SATA 6Gbps Barracuda XT drive spins 4x 500GB platters at 7200RPM with a big 64MB cache to prevent bottlenecks. It promises a sustained transfer rate of just 140MBps (compared to 600MBps / 4.8Gbps possible), MTBF of 750,000 hours, and carries a five-year warranty. The disk hits retail this week for about $299 list. Then you'll just need to find SATA 6G controller / MoBo to make the most of your new purchase -- fortunately, SATA 6Gbps is backward compatible with SATA 1.5Gbps or 3Gbps rigs until then. ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~
Hard disk technology (was: Disk based backup)
On Thu, Sep 17, 2009 at 4:10 PM, Alverson, Tom (Xetron) tom.alver...@ngc.com wrote: Modern hard drives can sustain, what, maybe 0.4 to 0.6 Gbit/sec? Even the 3 Gbit/sec we have now is much higher than that. How is moving to 6 Gbit/sec going to help? :) All they need to do is upgrade the (on-board) controllers to operate all the heads in parallel. For a 4 platter drive (8 heads) they could get an immediate 8X improvement in real read and write speeds. But that's got nothing to do with any particular revision of the SATA standard, right? You can interleave data across heads whether it's SATA, SCSI, or PATA. Even PATA hard drives have been presenting synthetic C/H/S geometry for something like a decade now, so it's not like the controllers don't already have to do the job. And given that *current* hard drives can't even keep up with SATA at 1.5 Gbit/sec, I would then ask: Why haven't they done this already? Or have they already, and the speeds we see today are *with* interleaving? (I'll ignore for the time being that most consumer hard disks these days only have one or two heads anyway.) -- Ben ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~
RE: Hard disk technology (was: Disk based backup)
I have to agree with Ben here. If it were easy to do, it would have been done already. I suspect the improved bus speeds will help devices that aren't current spinning disks (maybe flash based drives), or where we are able to present an array of disks at the end of the bus (e.g. external direct attached storage connected via eSATA) Cheers Ken -Original Message- From: Ben Scott [mailto:mailvor...@gmail.com] Sent: Friday, 18 September 2009 10:37 AM To: NT System Admin Issues Subject: Hard disk technology (was: Disk based backup) On Thu, Sep 17, 2009 at 4:10 PM, Alverson, Tom (Xetron) tom.alver...@ngc.com wrote: Modern hard drives can sustain, what, maybe 0.4 to 0.6 Gbit/sec? Even the 3 Gbit/sec we have now is much higher than that. How is moving to 6 Gbit/sec going to help? :) All they need to do is upgrade the (on-board) controllers to operate all the heads in parallel. For a 4 platter drive (8 heads) they could get an immediate 8X improvement in real read and write speeds. But that's got nothing to do with any particular revision of the SATA standard, right? You can interleave data across heads whether it's SATA, SCSI, or PATA. Even PATA hard drives have been presenting synthetic C/H/S geometry for something like a decade now, so it's not like the controllers don't already have to do the job. And given that *current* hard drives can't even keep up with SATA at 1.5 Gbit/sec, I would then ask: Why haven't they done this already? Or have they already, and the speeds we see today are *with* interleaving? (I'll ignore for the time being that most consumer hard disks these days only have one or two heads anyway.) -- Ben ~ Finally, powerful endpoint security that ISN'T a resource hog! ~ ~ http://www.sunbeltsoftware.com/Business/VIPRE-Enterprise/ ~