Both WD and Seagate are probably soiling themselves looking to the near future were SSDs will replace rotating hard drives in 80-90% of shipped PCs, think typical BestBuy or Dell computer not what Slashdotters buy. Flash/SSD lowers the cost of entry into the hard drive business, just look at all the companies making SSDs or announcing SSD. Somebody will make a chip that does SATA to Flash and then everybody and their dog will be turning out SSDs just like thumb drives. Then throw in that at least Intel and Samsung are both huge players in the flash market are you can start to see the problem. In 5 years WD and Seagate may be relegated to shipping niche "Bulk storage" drives and the the common boot drive will be sourced like video cards are now.
1) RPM. It is easier to spin a 2.5" platter at 15K than a 3.5" platter. (someone else can figure out the addtional energy but I would guess more than double the juice adduming uniform density.)
2) IOs per second. In large arrays the driving factor is not necessaraly throughput but IOs per second. Which leads to more transactions per second for your server farm. So more spindles = more IOs per second.
3) Access time. The bigger the drive the longer it takes the drive's processor to position the head. Therefore increasing access times. decreasing IO per second. I now its a trivial amount of time but it adds up over millions of IO.
4) Error correction. I cannot speak for IDE but each block on a SCSI drive has an Error Correction Code (ECC) which helps the drive recover from read errors. Again minimal.
5) Cynical answer. Smaller drives means your drive company sells more product to meet a given capacity.
educational point. SCSI is a protocol like IP or TCP. It can be tunneled through or carried by anything.
SPI -SCSI Parralel interface (old school).
FCP - Fibre channel protocol
SAS - Serial attached SCSI. SAS can also tunnel SATA.
iSCSI - scsi in TCP. (not ethernet)
SBP - SCSI Block Protocol. firewire.
ATAPI - yep SCSI ove IDE so your CDROM works.
many others.
The word "fair" is not being used to describe any "morality" of guerilla tactics. It is being used to describe parity of forces. No competent commander would ever take his forces into a "fair" fight if he/she can help it.
Re:Manufacturers are doing what they're supposed t
on
KISS
·
· Score: 1
The primary culprit here is probably the cheap EDTV (480p resolution) plasmas that Gateway and others sell. People think the have an HDTV because they shelled out $3000, but they do not. HDTV (720p or 1080i resolution) plasmas in general are >$5000. So in general buyer beware.
We ship PCs for work sometimes. We saved a couple of the boxes they came in and use those. This way you get all the foam inserts and stuff. If your packing material is gone maybe you can get some from the local apple store. Make sure everything is screwed back together before you box it up if you have had it open. The boxes vibrate alot and cards can come free if they are not screwed in. Finaly be sure to fill in the insurance section for the value of contents. Not the cheapest solution but it works.
Get several IDE adapters and run the cables out the back of the box. Use an another power suppply to spin the drives. I do not have any ideas about hot swapping. You could do a cheap environmental chamber with a cardboard box and no fans to see how the drives do without any ventilation. Then get iometer and write some tests. Be sure to do sevral passes with different byte patterns (00, AA, 55,FF) over the whole media. Also through in a large block of random accesses of varying both length and location. You should also do a butterfly pattern write read, FIRST LBA, LAST LBA, FIRST+1, LAST-1. Loop and let run as long as necessary to make you happy. The SCSI guys will do this kind of thing to drives for weeks-months non stop to figure MTBF and find other problems. They have specialized solutions for software but IOmeter should do unless you want to learn how to code direct disk accesses not file system.
Goto http://www.baen.com and try out the free library. They have several complete books online for reading or download. You can try out an author and then go get the rest of the series. It is kind of like drugs though, only the first part of a series is availble never the whole thing.
SATA 1.5 Gbps (not 1.2) GIGe ~1.06 Gbps Fibrechannel 2.125 Gbps Serial Attached SCSI 3.0 Gbps (Coming soon)
FYI SAS and SATA both share a physical layer and are roadmapped at 3,6,12 Gbps. 12 maybe hard but 6 is doable, maybe in 2006-7. SATA will lag behind mostly because of cost reasons, 3Gbps is possible today its just not cheap.
Only one device transmits on a parrallel bus at a time. (SCSI, PCI). Devices use some kind of HW based arbitration to decide whose turn it is. There are usually a couple of extra lines in the bus for this purpose. Traces on motherboards are the same length to prevent skew in the differential signals. A differential signal has a pair of signals and the value (1,0) is determined by the voltage difference. SATA, SAS, Fibrechannel, and GIGe are all 10b/8b serial protocols that use a differential pair to send a signal, so board traces must be identical for these serial transports also. For more info contact your local EE.
Many new drives these days use 2.5" platters. Less mass to spin at 15,000 rpm and therefore less power to do it. Also the standard form factor for drives will shrink to 2.5" from 3.5" in the next few years. So new math:
platters count will probably go down to 1 or 2. Again less juice needed. Currently the push in enterprise storage is towards more spindles not bigger drives. This provides more IO pers sec.
People keep trying to compare 1394 to ATA devices claiming it is a superior protocol, which it is. The problem with this is that all the features that make 1394 better make it vastly more expensive to produce. An ATA does about one thing get block(s) of data from the drive, one IO at a time. Lets call this a complexity of 1. 1394 gives you Multiple IOs, includes mutli initiator (complexity 2), different topologies (complexity 3). Multiple device types (complexity 4). The increase in complexity adds a geometric growth to the cost of engineering not a linear one. 1394 trully is competeing with parallel SCSI and Fibre Channel. Where it doesn't stand a chance, mostly because of industry inertia. Why waste time on 1394 when existing protocols do the job already.
Slashdot Mirror
Both WD and Seagate are probably soiling themselves looking to the near future were SSDs will replace rotating hard drives in 80-90% of shipped PCs, think typical BestBuy or Dell computer not what Slashdotters buy. Flash/SSD lowers the cost of entry into the hard drive business, just look at all the companies making SSDs or announcing SSD. Somebody will make a chip that does SATA to Flash and then everybody and their dog will be turning out SSDs just like thumb drives. Then throw in that at least Intel and Samsung are both huge players in the flash market are you can start to see the problem. In 5 years WD and Seagate may be relegated to shipping niche "Bulk storage" drives and the the common boot drive will be sourced like video cards are now.
I will try to avoid the SCSI vs IDE flame war.
1) RPM. It is easier to spin a 2.5" platter at 15K than a 3.5" platter. (someone else can figure out the addtional energy but I would guess more than double the juice adduming uniform density.)
2) IOs per second. In large arrays the driving factor is not necessaraly throughput but IOs per second. Which leads to more transactions per second for your server farm. So more spindles = more IOs per second.
3) Access time. The bigger the drive the longer it takes the drive's processor to position the head. Therefore increasing access times. decreasing IO per second. I now its a trivial amount of time but it adds up over millions of IO.
4) Error correction. I cannot speak for IDE but each block on a SCSI drive has an Error Correction Code (ECC) which helps the drive recover from read errors. Again minimal.
5) Cynical answer. Smaller drives means your drive company sells more product to meet a given capacity.
educational point. SCSI is a protocol like IP or TCP. It can be tunneled through or carried by anything.
SPI -SCSI Parralel interface (old school).
FCP - Fibre channel protocol
SAS - Serial attached SCSI. SAS can also tunnel SATA.
iSCSI - scsi in TCP. (not ethernet)
SBP - SCSI Block Protocol. firewire.
ATAPI - yep SCSI ove IDE so your CDROM works.
many others.
The word "fair" is not being used to describe any "morality" of guerilla tactics. It is being used to describe parity of forces. No competent commander would ever take his forces into a "fair" fight if he/she can help it.
The primary culprit here is probably the cheap EDTV (480p resolution) plasmas that Gateway and others sell. People think the have an HDTV because they shelled out $3000, but they do not. HDTV (720p or 1080i resolution) plasmas in general are >$5000. So in general buyer beware.
Yes return the game. I have returned games in the past. You may only get store credit though.
We ship PCs for work sometimes. We saved a couple of the boxes they came in and use those. This way you get all the foam inserts and stuff. If your packing material is gone maybe you can get some from the local apple store. Make sure everything is screwed back together before you box it up if you have had it open. The boxes vibrate alot and cards can come free if they are not screwed in. Finaly be sure to fill in the insurance section for the value of contents. Not the cheapest solution but it works.
Get several IDE adapters and run the cables out the back of the box. Use an another power suppply to spin the drives. I do not have any ideas about hot swapping. You could do a cheap environmental chamber with a cardboard box and no fans to see how the drives do without any ventilation. Then get iometer and write some tests. Be sure to do sevral passes with different byte patterns (00, AA, 55 ,FF) over the whole media. Also through in a large block of random accesses of varying both length and location. You should also do a butterfly pattern write read, FIRST LBA, LAST LBA, FIRST+1, LAST-1. Loop and let run as long as necessary to make you happy. The SCSI guys will do this kind of thing to drives for weeks-months non stop to figure MTBF and find other problems. They have specialized solutions for software but IOmeter should do unless you want to learn how to code direct disk accesses not file system.
Goto http://www.baen.com and try out the free library. They have several complete books online for reading or download. You can try out an author and then go get the rest of the series. It is kind of like drugs though, only the first part of a series is availble never the whole thing.
A few more and a correction.
SATA 1.5 Gbps (not 1.2)
GIGe ~1.06 Gbps
Fibrechannel 2.125 Gbps
Serial Attached SCSI 3.0 Gbps (Coming soon)
FYI
SAS and SATA both share a physical layer and are roadmapped at 3,6,12 Gbps. 12 maybe hard but 6 is doable, maybe in 2006-7. SATA will lag behind mostly because of cost reasons, 3Gbps is possible today its just not cheap.
Only one device transmits on a parrallel bus at a time. (SCSI, PCI). Devices use some kind of HW based arbitration to decide whose turn it is. There are usually a couple of extra lines in the bus for this purpose. Traces on motherboards are the same length to prevent skew in the differential signals.
A differential signal has a pair of signals and the value (1,0) is determined by the voltage difference. SATA, SAS, Fibrechannel, and GIGe are all 10b/8b serial protocols that use a differential pair to send a signal, so board traces must be identical for these serial transports also. For more info contact your local EE.
Hot Swap and room for 3.3v 5v and 12v.
Many new drives these days use 2.5" platters. Less mass to spin at 15,000 rpm and therefore less power to do it. Also the standard form factor for drives will shrink to 2.5" from 3.5" in the next few years.
So new math:
4.908 - 0.196 = 4.712 in^2
4.712 *50/8= 29.45 TB per platter
platters count will probably go down to 1 or 2. Again less juice needed. Currently the push in enterprise storage is towards more spindles not bigger drives. This provides more IO pers sec.
People keep trying to compare 1394 to ATA devices claiming it is a superior protocol, which it is. The problem with this is that all the features that make 1394 better make it vastly more
expensive to produce. An ATA does about one thing get block(s) of data from the drive, one IO at a time. Lets call this a complexity of 1. 1394 gives you Multiple IOs, includes mutli initiator (complexity 2), different topologies (complexity 3). Multiple device types (complexity 4). The increase in complexity adds a geometric growth to the cost of engineering not a linear one. 1394 trully is competeing with parallel SCSI and Fibre Channel. Where it doesn't stand a chance, mostly because of industry inertia. Why waste time on 1394 when existing protocols do the job already.
To my knowledge SSA has generally been supplanted by Fibre Channel. IBM maintains current installs but probably does not do new ones.