10k hard drives, and hard drive evolution

S

Sovereign

2[H]4U
Joined
Mar 21, 2005
Messages
3,098
I believe there was a time (before I got into computers) that 5400RPM was the standard and 7200RPM was "fast" (correct me if I'm wrong). Is there a chance that consumer computers (or gaming computers at least) will move mainly to 10k in your opinoin? And is there any HDD faster than 15k? I heard something about 12.5k somewhere but don't think it was substantial.
 
I'd think 10k has a chance.. look how long we've had 10k drives and how little they cost now. 36g raptor for $75 ish. Now, it may be a couple of years. but it will happen.
 
Cache. HD's went from kb's to 2MB to 8MB and now 16MB cache buffers. 16MB cache paired with 7200RPM's usually beats out 10,000RPM with 8MB cache right now.
 
You can't argue with the rotational latency of a 10k drive vs a 7.2k rpm. Cache can only mask the delay. 10k with the same ammount of cache will always beat 7.2k with the same cache.

10k is the future.
 
You can get most of 10K level seek performance out of a 7200RPM drive by simply putting a 10K level actuator on a 7200RPM spindle. Think about it.

Typical 7200RPM drive seek time: ~9.0ms average seek (actuator) + 4.2ms latency (spindle) = 13.2ms
Typical 10K drive seek time: ~5.0ms average seek (actuator) + 3.0ms latency = 8.0ms

If you compare our 10K drive to the 7200RPM drive above, you'll see that the 7200RPM drive has a 65% higher read service time than the 10K drive (13.2-8.0)/8.0=.65. Let's put that 10K grade actuator on the 7200RPM drive and see what happens:

5.0ms seek (actuator) + 4.2ms latency (spindle) = 9.2ms
Same 8.0ms for the 10K drive.

(9.2-8.0)/8.0=.15 Now the 7200RPM drive only takes 15% longer than the 10K drive.

Given ~9.0ms seeks and 4.2ms latency on a 7200RPM drive, and we want to increase mechanical performance, what's the best way to go about it?

Well, we can increase spindle speed ~40% (Going from 7200 to 10K is an increase of 38.8%). That makes seeks 1.2ms faster. 1.2ms faster...what would it take to get those 1.2ms out of the actuator?

9.0-1.2=7.8
1.2/9.0=.1333, about 15%

So, increasing actuator power and precision by 15% has the same effect on performance as increasing spindle speed by 40%. Now I'm not an engineer, and I certainly don't design hard drives. However, if I'm faced with raising something by 40% or lowering something by 15% to achieve a given goal, I'm pretty confident that the 15% will be easier to achieve...it's less sweeping adjustment.
 
It would seem that way, but just ask yourself why do all the nice, server level drives spin at 15k or 10k... There is no substitute for rpm.
 
Because enterprises pay to play. $100 is the absolute minimum for a SCSI HDD, compared to $40-50 in the ATA world. High end is at a similar premium, with 300GB 10K and 147GB 15K units fetching ~$800, ~$1000 if you believe in paying the Seagate tax :eek:

Also, high RPMs help...but has there been a significant increase in enterprise drive performance since the Fujitsu MAS debuted in 2003, over three years ago?

Are mechnically superior drives automatically faster on the desktop?

Is the Raptor outselling anything?

Are enthusiasts flocking to SAS in droves?

Will you pay over $100 for a 36GB drive? How about $350-400 for a 147GB drive?

Have ATA drives increased significantly in performance in the same time frame, thanks to increased density, improved firmware, etc? Absolutely.
 
You must log in or register to reply here.
Back
Top