Advice Please: The Importance of Hard Drive RPMs

kony wrote:

On Thu, 19 Aug 2004 04:48:06 GMT, CJT
wrote:



You must be on crack. It is MUCH, MUCH, MUCH MUCH slower that
anyone with semi-modern gear is accustomed to. Even GbE on a PC
is slower than budget local storage HDD. Server-side apps are a
logistic solution, performance be damned.


You're just pulling things out of your *ss. I'm telling you about
actual experience.



Oops, I guess I forgot that I've never used a computer before,
LOL.




Of course, if you just have one (truly "personal") computer, it'll have
a disk attached.

But I think for many people, disk drive speed is pretty low on the list
of things on which they should be spending money.



That's why some people buy newer computer then soon feel it isn't
much faster, because they didn't significantly improve the
bottleneck to their use, which is often the HDD.

Not in my experience for typical office tasks in a properly
configured system.


There is no such thing as "properly configured" that will change
the fact that data I/O is significantly slower over a LAN,
compared to any modern HDD. No grand theory changes that, all
you have is additional overhead in the already-slowest part of a
system, at least for these light tasks.



The nice thing about I/O is that not much of it is needed for
a typical user. It's easy enough to judge for yourself -- just
watch your disk activity light and see what fraction of the time
it's on. When that LED is not lit, it doesn't make a bit of
difference how fast the disk is spinning, or even if it's spinning
at all.

If you're doing video editing, or some games, you will benefit
from a fast disk. If you're browsing the Web, listening to music,
word processing, or using a spreadsheet, you probably aren't
hitting the disk much, so the number of seconds per day spent
waiting for the disk is probably quite small.

--
The e-mail address in our reply-to line is reversed in an attempt to
minimize spam. Our true address is of the form .

kony wrote:

On Thu, 19 Aug 2004 04:45:55 GMT, CJT
wrote:



Nonsense, everything the system is running is loaded from HDD,
and the speed is directly effected by speed of that hard drive.
A Celeron 800 with a WD Raptor HDD will feel faster for everday
use than a P4 3.2GHz with a budget-grade 40GB HDD.

Nothing wrong with choosing cheapest GB/$ for mass storage, but
it cripples a system to use such drives as primary app or OS
drive.

I disagree. If you spend all day browsing and word processing, you
load your browser and word processor once in the morning, and once
they're open then opening them isn't any longer an issue (unless
your machine crashes a lot -- but that's not usually the disk's fault).


I suppose you're just a troll, since I already told you that the
browser caches all those files to the HDD when "browsing".

The actual amount of data involved in caching web pages is quite small.

No matter how much you disagree, time and time again people
everywhere notice the difference between an old/slow HDD and
something modern/fast, not to mention benchmarks. A lot of
memory will reduce need for HDD access, but the two are
complimentary storage, not one a replacement for the other.

I guess we'll just have to disagree. I've stated my position.
If it makes you feel good to have a faster disk, and you've
convinced yourself you can detect the difference, that's ok
with me. There are people who claim they can detect the
difference when they change power cords on their stereos, too.

Watch your disk activity light. If it's on a lot, you might
benefit from a faster disk. If it hardly ever blinks, there's
little harm buying a faster disk (except possibly $$), but you're
deluding yourself if you think it's going to make a big difference
in your ability to get things done.

If benchmarks are your "thing" then go for it.

--
The e-mail address in our reply-to line is reversed in an attempt to
minimize spam. Our true address is of the form .

JAD wrote:

Makes little difference unless


Your a benchmark watcher

your transferring huge files between partitions..and maybe drives on a
regular basis, but there are other factors that effect data speed
between 2 physical drives other than HD RPM speed

Your using something other than OB IDE...SCSI drives (and sata, I
think- just started looking at moving to this format and can't give
you an opinion yet) would benefit or if your running a separate IDE
controller

If your running a file server....even this is not necessary.....lots
of memory can make up easily for HD rpm. I would look for reliability,
warranty length, and if you want, the buffer size ( I don't see/feel
that its any faster 4 or 8)along with a good sale price

all my opinion, works for me sale gimmicks are every where

Indeed. Disk manufacturers do what they can to distinguish
their drives from those of their competitors. It's hard to
prove reliability, so they focus on speed. And speed isn't
bad -- it's just not necessarily worth a premium for small
performance increases once a "fast enough" threshold is passed.

If 5% of your time is spent waiting for your disk drive (and
I doubt it's even that high for many computer users), then
doubling its performance will only make about a 2.5% difference
in your overall throughput. Some people could save 2.5% of
their time by simply not worrying about such issues. g

Personally, I'd rather have slightly slower drives that were
cheaper, quieter and used less power (i.e. that generated less
heat and cost less to run, as well as perhaps being more reliable).
Where performance is an issue, RAID can be used, and will likely
make a much bigger difference than increasing spindle RPM.



"Darren Harris" wrote in message
om...

Can anyone tell me if hard drive spindle speed is an important

factor

to consider when purchasing a hard drive?

Or should I just concentrate on average latency, average access, and
max. full seek time?

I ask because two hard drives with a data rate of 80mps can differ

in

these other respects.

Thanks a lot.

Darren Harris
Staten Island, New York.





--
The e-mail address in our reply-to line is reversed in an attempt to
minimize spam. Our true address is of the form .

CJT wrote:

JAD wrote:

Makes little difference unless


Your a benchmark watcher

your transferring huge files between partitions..and maybe drives on a
regular basis, but there are other factors that effect data speed
between 2 physical drives other than HD RPM speed

Your using something other than OB IDE...SCSI drives (and sata, I
think- just started looking at moving to this format and can't give
you an opinion yet) would benefit or if your running a separate IDE
controller

If your running a file server....even this is not necessary.....lots
of memory can make up easily for HD rpm. I would look for reliability,
warranty length, and if you want, the buffer size ( I don't see/feel
that its any faster 4 or 8)along with a good sale price

all my opinion, works for me sale gimmicks are every where


Indeed. Disk manufacturers do what they can to distinguish
their drives from those of their competitors. It's hard to
prove reliability, so they focus on speed. And speed isn't
bad -- it's just not necessarily worth a premium for small
performance increases once a "fast enough" threshold is passed.

If 5% of your time is spent waiting for your disk drive (and
I doubt it's even that high for many computer users), then
doubling its performance will only make about a 2.5% difference
in your overall throughput. Some people could save 2.5% of
their time by simply not worrying about such issues. g

Personally, I'd rather have slightly slower drives that were
cheaper, quieter and used less power (i.e. that generated less
heat and cost less to run, as well as perhaps being more reliable).
Where performance is an issue, RAID can be used,

Which just shot to hell the "less heat, quieter, cost less, and more
reliability" argument.

and will likely
make a much bigger difference than increasing spindle RPM.

The theory would suggest so but it just doesn't manifest itself as well in
the real world use of it.

Now, to be fair about it, when people see the 'difference' with a 7200 vs a
5400 it's usually combined with a rather dramatic increase in capacity,
which usually means a density increase, and, if it's linear density, that
improves the performance as well, but that part is not due to the 'RPM'.


"Darren Harris" wrote in message
om...

Can anyone tell me if hard drive spindle speed is an important


factor

to consider when purchasing a hard drive?

Or should I just concentrate on average latency, average access, and
max. full seek time?

I ask because two hard drives with a data rate of 80mps can differ


in

these other respects.

Thanks a lot.

Darren Harris
Staten Island, New York.

David Maynard wrote:
CJT wrote:

JAD wrote:

Makes little difference unless


Your a benchmark watcher

your transferring huge files between partitions..and maybe drives on a
regular basis, but there are other factors that effect data speed
between 2 physical drives other than HD RPM speed

Your using something other than OB IDE...SCSI drives (and sata, I
think- just started looking at moving to this format and can't give
you an opinion yet) would benefit or if your running a separate IDE
controller

If your running a file server....even this is not necessary.....lots
of memory can make up easily for HD rpm. I would look for reliability,
warranty length, and if you want, the buffer size ( I don't see/feel
that its any faster 4 or 8)along with a good sale price

all my opinion, works for me sale gimmicks are every where



Indeed. Disk manufacturers do what they can to distinguish
their drives from those of their competitors. It's hard to
prove reliability, so they focus on speed. And speed isn't
bad -- it's just not necessarily worth a premium for small
performance increases once a "fast enough" threshold is passed.

If 5% of your time is spent waiting for your disk drive (and
I doubt it's even that high for many computer users), then
doubling its performance will only make about a 2.5% difference
in your overall throughput. Some people could save 2.5% of
their time by simply not worrying about such issues. g

Personally, I'd rather have slightly slower drives that were
cheaper, quieter and used less power (i.e. that generated less
heat and cost less to run, as well as perhaps being more reliable).
Where performance is an issue, RAID can be used,


Which just shot to hell the "less heat, quieter, cost less, and more
reliability" argument.

Huh? Build your RAID of such drives.


and will likely
make a much bigger difference than increasing spindle RPM.


The theory would suggest so but it just doesn't manifest itself as well
in the real world use of it.

Gee, maybe you should tell EMC.

Now, to be fair about it, when people see the 'difference' with a 7200
vs a 5400 it's usually combined with a rather dramatic increase in
capacity, which usually means a density increase, and, if it's linear
density, that improves the performance as well, but that part is not due
to the 'RPM'.


Probably right.


"Darren Harris" wrote in message
om...

Can anyone tell me if hard drive spindle speed is an important



factor

to consider when purchasing a hard drive?

Or should I just concentrate on average latency, average access, and
max. full seek time?

I ask because two hard drives with a data rate of 80mps can differ



in

these other respects.

Thanks a lot.

Darren Harris
Staten Island, New York.










--
The e-mail address in our reply-to line is reversed in an attempt to
minimize spam. Our true address is of the form .

David Maynard wrote:

CJT wrote:

David Maynard wrote:

CJT wrote:

kony wrote:

On Wed, 18 Aug 2004 21:29:03 GMT, CJT
wrote:



Alright, I'll concede there, but a 7200 will still provide a
very noticable performance increase over a 5400. Not just in
drive benchmarks, but in day to day computer usage.





I think the devil will be in the details. If you mostly just
browse the
Web, I doubt your disk will be exercised much. If you do a lot of
video
editing, you probably want something pretty fast -- most likely RAID.
There's a whole range in between (and perhaps beyond).






Clueless.

Browser caches everything to disk, and reloads it all from this
cache until pages are refreshed unless brower is changed from
defaults.

The only time HDD speed doesn't matter much is when system has A)
Excess system memory to cache files B) Limited multitasking so
files are never flushed from this cache.





I stand by what I said. Watch your disk light some time. If it's
on a lot, then the disk speed might make a difference. If it hardly
ever flashes, then your drive's speed doesn't matter one bit.


Most folk's applications have a heck of a time getting from the hard
drive to RAM without flashing the LED.


No disagreement here, but for most folks that's a minuscule fraction of
the time they spend sitting in front of the computer.


Well, we could quibble over what 'minuscule' means in this context but
the reality of it is that most home users don't load up Word and then
leave it there all day while they, however frequently or infrequently,
pound out documents; they, e.g. families, are often a competing set of
users with applications going up and down rather often and even a
'single' gamer doesn't necessarily load up just one game for the day.
And it really doesn't matter if 'mathematically' the disk usage is a
'small percentage' of the total time because what a user 'feels' and
gauges things by is how long it takes between 'click-click' and whatever
they expect to happen from it.

And that's before we even get to doing a couple of things simultaneously
and/or burning CD/DVDs, playing videos/MP3s, etc.

Even playing DVD images doesn't tax modern "slow" drives. Video
_editing_ might, but there you're trying to go (much) faster than
real time. I can play half a dozen .wav files (which are more data
intensive than MP3s) simultaneously and the disk LED hardly lights.

If you're playing uncompressed video, then I can understand why you
might want a fast drive. But that's not a very smart thing to do.


I have yet to see a user who didn't notice the difference between a 15
gig 5400 RPM drive and a 120 gig 7200 RPM drive. The later is simply
faster.



--
The e-mail address in our reply-to line is reversed in an attempt to
minimize spam. Our true address is of the form .

CJT wrote:

David Maynard wrote:

CJT wrote:

JAD wrote:

Makes little difference unless


Your a benchmark watcher

your transferring huge files between partitions..and maybe drives on a
regular basis, but there are other factors that effect data speed
between 2 physical drives other than HD RPM speed

Your using something other than OB IDE...SCSI drives (and sata, I
think- just started looking at moving to this format and can't give
you an opinion yet) would benefit or if your running a separate IDE
controller

If your running a file server....even this is not necessary.....lots
of memory can make up easily for HD rpm. I would look for reliability,
warranty length, and if you want, the buffer size ( I don't see/feel
that its any faster 4 or 8)along with a good sale price

all my opinion, works for me sale gimmicks are every where




Indeed. Disk manufacturers do what they can to distinguish
their drives from those of their competitors. It's hard to
prove reliability, so they focus on speed. And speed isn't
bad -- it's just not necessarily worth a premium for small
performance increases once a "fast enough" threshold is passed.

If 5% of your time is spent waiting for your disk drive (and
I doubt it's even that high for many computer users), then
doubling its performance will only make about a 2.5% difference
in your overall throughput. Some people could save 2.5% of
their time by simply not worrying about such issues. g

Personally, I'd rather have slightly slower drives that were
cheaper, quieter and used less power (i.e. that generated less
heat and cost less to run, as well as perhaps being more reliable).
Where performance is an issue, RAID can be used,



Which just shot to hell the "less heat, quieter, cost less, and more
reliability" argument.


Huh? Build your RAID of such drives.

Two "less heat, quieter, cost less, and more reliable" drives raided
together to make up for the performance loss are no longer "less heat,
quieter, cost less, and more reliable" than the single faster one they're
replacing.

If you want performance then your best first bet is to go to the inherently
faster drive because that will give you the lion's share of the increase
with a "less heat, quieter, cost less, and more reliable" solution.

and will likely
make a much bigger difference than increasing spindle RPM.



The theory would suggest so but it just doesn't manifest itself as
well in the real world use of it.


Gee, maybe you should tell EMC.

Surely they already know you don't get twice the performance with a pair of
raided drives.


Now, to be fair about it, when people see the 'difference' with a 7200
vs a 5400 it's usually combined with a rather dramatic increase in
capacity, which usually means a density increase, and, if it's linear
density, that improves the performance as well, but that part is not
due to the 'RPM'.


Probably right.

CJT wrote:

David Maynard wrote:

CJT wrote:

David Maynard wrote:

CJT wrote:

kony wrote:

On Wed, 18 Aug 2004 21:29:03 GMT, CJT
wrote:



Alright, I'll concede there, but a 7200 will still provide a
very noticable performance increase over a 5400. Not just in
drive benchmarks, but in day to day computer usage.






I think the devil will be in the details. If you mostly just
browse the
Web, I doubt your disk will be exercised much. If you do a lot
of video
editing, you probably want something pretty fast -- most likely
RAID.
There's a whole range in between (and perhaps beyond).







Clueless.

Browser caches everything to disk, and reloads it all from this
cache until pages are refreshed unless brower is changed from
defaults.

The only time HDD speed doesn't matter much is when system has A)
Excess system memory to cache files B) Limited multitasking so
files are never flushed from this cache.






I stand by what I said. Watch your disk light some time. If it's
on a lot, then the disk speed might make a difference. If it hardly
ever flashes, then your drive's speed doesn't matter one bit.


Most folk's applications have a heck of a time getting from the hard
drive to RAM without flashing the LED.


No disagreement here, but for most folks that's a minuscule fraction of
the time they spend sitting in front of the computer.


Well, we could quibble over what 'minuscule' means in this context but
the reality of it is that most home users don't load up Word and then
leave it there all day while they, however frequently or infrequently,
pound out documents; they, e.g. families, are often a competing set of
users with applications going up and down rather often and even a
'single' gamer doesn't necessarily load up just one game for the day.
And it really doesn't matter if 'mathematically' the disk usage is a
'small percentage' of the total time because what a user 'feels' and
gauges things by is how long it takes between 'click-click' and
whatever they expect to happen from it.

And that's before we even get to doing a couple of things
simultaneously and/or burning CD/DVDs, playing videos/MP3s, etc.


Even playing DVD images doesn't tax modern "slow" drives. Video
_editing_ might, but there you're trying to go (much) faster than
real time. I can play half a dozen .wav files (which are more data
intensive than MP3s) simultaneously and the disk LED hardly lights.

If you're playing uncompressed video, then I can understand why you
might want a fast drive. But that's not a very smart thing to do.

I didn't say any one thing would 'tax' the slower drive. I simply said
there would be a noticeable improvement with the faster one.


I have yet to see a user who didn't notice the difference between a 15
gig 5400 RPM drive and a 120 gig 7200 RPM drive. The later is simply
faster.

David Maynard wrote:

CJT wrote:

David Maynard wrote:

CJT wrote:

JAD wrote:

Makes little difference unless


Your a benchmark watcher

your transferring huge files between partitions..and maybe drives on a
regular basis, but there are other factors that effect data speed
between 2 physical drives other than HD RPM speed

Your using something other than OB IDE...SCSI drives (and sata, I
think- just started looking at moving to this format and can't give
you an opinion yet) would benefit or if your running a separate IDE
controller

If your running a file server....even this is not necessary.....lots
of memory can make up easily for HD rpm. I would look for reliability,
warranty length, and if you want, the buffer size ( I don't see/feel
that its any faster 4 or 8)along with a good sale price

all my opinion, works for me sale gimmicks are every where





Indeed. Disk manufacturers do what they can to distinguish
their drives from those of their competitors. It's hard to
prove reliability, so they focus on speed. And speed isn't
bad -- it's just not necessarily worth a premium for small
performance increases once a "fast enough" threshold is passed.

If 5% of your time is spent waiting for your disk drive (and
I doubt it's even that high for many computer users), then
doubling its performance will only make about a 2.5% difference
in your overall throughput. Some people could save 2.5% of
their time by simply not worrying about such issues. g

Personally, I'd rather have slightly slower drives that were
cheaper, quieter and used less power (i.e. that generated less
heat and cost less to run, as well as perhaps being more reliable).
Where performance is an issue, RAID can be used,




Which just shot to hell the "less heat, quieter, cost less, and more
reliability" argument.



Huh? Build your RAID of such drives.


Two "less heat, quieter, cost less, and more reliable" drives raided
together to make up for the performance loss are no longer "less heat,
quieter, cost less, and more reliable" than the single faster one
they're replacing.

A RAID installation can be much faster than any drive you can buy, so
the comparison is an empty one.


If you want performance then your best first bet is to go to the
inherently faster drive because that will give you the lion's share of
the increase with a "less heat, quieter, cost less, and more reliable"
solution.

If you can do it with a fast drive, but not a less fast one, then ok.
But that's a fairly narrow band of applicability.


and will likely
make a much bigger difference than increasing spindle RPM.




The theory would suggest so but it just doesn't manifest itself as
well in the real world use of it.



Gee, maybe you should tell EMC.


Surely they already know you don't get twice the performance with a pair
of raided drives.

They also know 10000 RPMs aren't double 5400.


Now, to be fair about it, when people see the 'difference' with a
7200 vs a 5400 it's usually combined with a rather dramatic increase
in capacity, which usually means a density increase, and, if it's
linear density, that improves the performance as well, but that part
is not due to the 'RPM'.


Probably right.





--
The e-mail address in our reply-to line is reversed in an attempt to
minimize spam. Our true address is of the form .

On Thu, 19 Aug 2004 07:38:26 GMT, CJT
wrote:



I suppose you're just a troll, since I already told you that the
browser caches all those files to the HDD when "browsing".

The actual amount of data involved in caching web pages is quite small.

Untrue, or at least only relatively small. I rebooted the system
I'm on right now, a few hours ago... since then the empty cache
has gained over 2000 files. a little over 20MB. It would've been
even larger but I have about 70% of the ad host servers and
shockwave flash blocked.

Point being, this is still a theory (that HDD speed isn't
signficant) that is far enough off that even a user with naked
eye can see the performance difference.



No matter how much you disagree, time and time again people
everywhere notice the difference between an old/slow HDD and
something modern/fast, not to mention benchmarks. A lot of
memory will reduce need for HDD access, but the two are
complimentary storage, not one a replacement for the other.

I guess we'll just have to disagree. I've stated my position.
If it makes you feel good to have a faster disk, and you've
convinced yourself you can detect the difference, that's ok
with me. There are people who claim they can detect the
difference when they change power cords on their stereos, too.

Ok, i can accept that we'll just disagree, and be glad that my
systems are all faster because of it.



Watch your disk activity light. If it's on a lot, you might
benefit from a faster disk. If it hardly ever blinks, there's
little harm buying a faster disk (except possibly $$), but you're
deluding yourself if you think it's going to make a big difference
in your ability to get things done.

If benchmarks are your "thing" then go for it.

No need to benchmark, it is plain as the nose on your face that
HDD speed is a primary bottleneck for many basic PC uses. It
matters more than CPU speed, FSB speed, memory bus speed, for a
lot of tasks. If you want to argue that slower HDD only causes a
few seconds additional wait, well sure that's true, a few seconds
over and over again, which is clearly not desired else we'd have
stuck with 80486 boxes instead of upgrading every so often.