Does 15k SCSI justify the price over 10k SATA?

[Ge|atinousFury]

Title says it all.

Over the past few days I've been looking at raptors, and the SCSI 15k drives are catching my eye. I have no experience with SCSI but I know that a drive, controller, cable, and terminator are required, but I'd have no idea what to add to my cart.

1. How much "everyday noticeable" speed increase will I notice from a 15k rpm SCSI vs. 10k rpm Raptor?

2. Let's say that theoretically I wanted to buy a 15k SCSI drive for use as my main Windows drive and use my IDE drive for backup/storage.

Can someone please make me out a sample preliminary list of what products I might need to buy (specific model numbers would be very nice and appreciated) to get this drive going?

[donny_paycheck]

Hugely dependent upon application, but I'm going to just say no.

I have a 73.4gb Maxtor Atlas 15K ($470) on a LSI U160 ($35), with a 2-device terminated cable ($15). Yeah, it's fast, and yeah, faster than a 74 gig Raptor is, but not proportional to the price. I would not recommend paying twice for this setup, especially not if you have an ICH5 motherboard with native SATA support. (I know you don't, but others reading this may)

I can post some benchmarks if you want but the best place to look is at http://storagereview.com/comparison.html

For quickness, click on this link. It's a head-to-head gaming benchmark comparison of the 73.4 Atlas 15k and 74 Raptor: http://storagereview.com/php/benchmark/compare_rtg_2001.php?typeID=10&testbedID=3&osID=4&raidconfigID=1&numDrives=1&devID_0=232&devID_1=254&devCnt=2

[Ge|atinousFury]

Originally posted by donny_paycheck
Hugely dependent upon application, but I'm going to just say no.

I have a 73.4gb Maxtor Atlas 15K ($470) on a LSI U160 ($35), with a 2-device terminated cable ($15). Yeah, it's fast, and yeah, faster than a 74 gig Raptor is, but not proportional to the price. I would not recommend paying twice for this setup, especially not if you have an ICH5 motherboard with native SATA support. (I know you don't, but others reading this may)

I can post some benchmarks if you want but the best place to look is at http://storagereview.com/comparison.html

For quickness, click on this link. It's a head-to-head gaming benchmark comparison of the 73.4 Atlas 15k and 74 Raptor: http://storagereview.com/php/benchmark/compare_rtg_2001.php?typeID=10&testbedID=3&osID=4&raidconfigID=1&numDrives=1&devID_0=232&devID_1=254&devCnt=2

If you don't mind me asking, what made you go for the SCSI then? Were you hoping for a more noticeable difference?

[donny_paycheck]

Originally posted by Ge|atinousFury
If you don't mind me asking, what made you go for the SCSI then? Were you hoping for a more noticeable difference?
Well I already had the U160 card, no SATA capability (although I do now), and I was willing to pay the extra cash. Personally I probably would again. But from a cost/performance ratio standpoint it would not be prudent.

[Ge|atinousFury]

Originally posted by donny_paycheck

Well I already had the U160 card, no SATA capability (although I do now), and I was willing to pay the extra cash. Personally I probably would again. But from a cost/performance ratio standpoint it would not be prudent.

Well that's understandable. I'm thinking I don't actually NEED any more than 36.7GB for Windows/programs but I'm seeing on the storagereview graphs that the 74GB Raptor surpassed even a 15k SCSI drive...that's impressive!

Edit: I do realize that it just surpassed that specific 15k drive in that specific application but it still warrants an eyebrow raise..

[Th0r]

I would go for the Raptor, the SCSI drive doesn't offer any big jump n performance at double the price...

[gustav]

I believe SCSI are best for server applications. where they may be put under a lot of stress for long periods of time. servers run differently than pc's and so the hard drives made for servers, prolly wouldnt show much difference over a 74gb raptor.

i have a 74 gb raptor now, i used to have a 36.7. i dont notice much difference between the two besides that the 74gb is much quieter. it still makes a racket when reading or writing, but at idle spin, it doesnt make that high pitch whining sound. the only reason i went for the 74gb raptor is cuz i couldnt get two 36.7 raptors into RAID 0.

[Ge|atinousFury]

Originally posted by gustav
I believe SCSI are best for server applications. where they may be put under a lot of stress for long periods of time. servers run differently than pc's and so the hard drives made for servers, prolly wouldnt show much difference over a 74gb raptor.

i have a 74 gb raptor now, i used to have a 36.7. i dont notice much difference between the two besides that the 74gb is much quieter. it still makes a racket when reading or writing, but at idle spin, it doesnt make that high pitch whining sound. the only reason i went for the 74gb raptor is cuz i couldnt get two 36.7 raptors into RAID 0.

I wonder if 2 36.7GB Raptors in RAID 0 would be faster than 1 74GB Raptor in benchmarks and the like...


I also wonder why the 15k SCSI drives don't annihilate the 10k Raptors in the benchmarks....logically one can assume that since the test results don't show the 15k SCSI drive's performance improvement proportional to the rpm improvement, then HDD performance depends on more than just revolutions per minute. Is this correct?

[NovaShine]

The 36GB raptors in raid 0 would completely waste a 74GB raptor alone.

Also, the raptors were made to run in server conditions, so a pretty long life expectancy (several years i think) continuous run at 100% load.

[Ge|atinousFury]

Originally posted by NovaShine
The 36GB raptors in raid 0 would completely waste a 74GB raptor alone.



Well, now to just decide whether or not I only need 36.7GB for Windows/programs or 74GB

[donny_paycheck]

Originally posted by NovaShine
The 36GB raptors in raid 0 would completely waste a 74GB raptor alone.
Depends on what benchmark you're running. For throughput, most definitely. However, seek time tends to increase slightly in RAIDs due to phasing disparities between the drives. In most cases, seek time and burst transfer rates are what determine a majority of I/O performance.

The gain of going from a single drive to a stripe set of the same is still there, it's just not as significant as the doubling in throughput that ATTO will detect. This is why benchmarks, and especially disk benchmarks in particular, should be regarded critically unless you are 100% sure they stress exactly the aspect you will be using the hardware for.

[Ge|atinousFury]

Originally posted by donny_paycheck

Depends on what benchmark you're running. For throughput, most definitely. However, seek time tends to increase slightly in RAIDs due to phasing disparities between the drives. In most cases, seek time and burst transfer rates are what determine a majority of I/O performance.

The gain of going from a single drive to a stripe set of the same is still there, it's just not as significant as the doubling in throughput that ATTO will detect. This is why benchmarks, and especially disk benchmarks in particular, should be regarded critically unless you are 100% sure they stress exactly the aspect you will be using the hardware for.

Is throughput another name for "transfer rate"?

[donny_paycheck]

Originally posted by Ge|atinousFury
Is throughput another name for "transfer rate"?
Yeah. It's how much you can "put through" a medium in a given unit of time.

[Th0r]

No, i want to see 2x74 Raptors in a RAID... LoL...

[gustav]

you got $520 for hard drives, go ahead and do it and let us know :D :p