How do SCSI drives get such low seek times?

[astroview]

Is it because they are so fast, ie. the 3.9 ms seek time on the 15,000 rpm Seagate?

Ahh, when will IDE drives have such nice seek times! :Q

 

[SCUBA]

3.9??
on hdtach it wont be better than 5ms

 

[Radboy]

it's all in the actuators, which position the read/write heads.

the new x15-36lp's have faster seeks than the (1st gen) x15s.

seek times have nothing to do with spindle speed.

storagereview has much info on the subject.

 

[Pariah]

The platters are smaller, which will help quite a bit.

 

[andalas]

it's easy
it spins at 15.000 RPM that's why it's faster.

IDE drive have more platter density than most scsi.
higher platter density also will decrease seek time.

 

[andalas]

<< it's all in the actuators, which position the read/write heads.

the new x15-36lp's have faster seeks than the (1st gen) x15s.

seek times have nothing to do with spindle speed.

storagereview has much info on the subject. >>



The 2nd generation X15 have higher platter density
that's why it decrease the seek time plus some other
improvement.

 

[Sugadaddy]

What I want to know is why can't those geniuses working for the hard drive manufacturers come up with an IDE drive faster than 7200 rpm?? At least give us 10 000 rpm, that would be a start...

 

[Noriaki]

<< What I want to know is why can't those geniuses working for the hard drive manufacturers come up with an IDE drive faster than 7200 rpm?? At least give us 10 000 rpm, that would be a start... >>

They very easily could, but a 10,000rpm IDE drive would undercut 10,000rpm SCSI drive sales. When 15k is standard for SCSI and 18,000 RPMs are the high end IDE will move up to 10k RPM.


As to the seek times, well the spindle speed lowers rotational latency which is part of overall access time, but the actual seek time is dependant on the actuators. IDE drives could have actuators as good as SCSI drives, but again, the big boys don't want to undercut there top ends.

Why doesn't a Camaro have the HP and Handling of a Corvette? Because GM would sell a lot less Vettes.

Same thing here.

 

[Pariah]

Increased RPM's does not reduce seek time, only latency and access time. Increased areal density does not have any real affect on seek time either, though theoretically it could/should.

Drive manufacturer's could pretty easily slap an IDE interface on an X15-36LP, it's not a case of not being able to, it's not wanting to. There is no market for such drives, so they aren't produced.

 

[andalas]

I think you should see some 10K IDE sometimes in the future.
The barrier of course is price and heat.

If you take a look at SCSI price you would see that the price
difference between 7.2K and 10K is &quot;almost&quot; two times.

If you have 10K-15K scsi you should invest in HD.
I just don't know if many IDE users really want to invest
in a lot of accessories just to add a HD.
That's just my thought and it's not a fact.

 

[MplsBob]

I may stand corrected on this-

I was under the impression:

(1) That seek time was the time to move from the current track to the desired track.

(2) That access time was the seek time plus the time required (called rotational latency) for the disk to rotate the desired record under the head.

(3) That both of these values are usually stated in term of averages.

(4) That average seek would be the time required to move the head(s) from the far inner track to the far outer track, with the result being divided by 2 to form the average.

(5) That the average rotational latency is equal to the time required for one rotation of the disk divided by 2 to form the average.

Having said this, seek time would directly relate to how quickly the head(s) can be re-positioned. Also, rotational speed (rpm) would have nothing what-so-ever to do with seek time.

Finally, the desired goal would be to reduce both of these values as much as possible.

Please feel free to correct me if I am wrong!

===========
As for myself, I would like someone to make a cheap multi-gigabyte solid state disk with virtually instantaneous access.

 

[Noriaki]

<< (1) That seek time was the time to move from the current track to the desired track.

(2) That access time was the seek time plus the time required (called rotational latency) for the disk to rotate the desired record under the head.

(3) That both of these values are usually stated in term of averages.

(4) That average seek would be the time required to move the head(s) from the far inner track to the far outer track, with the result being divided by 2 to form the average.

(5) That the average rotational latency is equal to the time required for one rotation of the disk divided by 2 to form the average. >>



You are correct on all counts I believe.