5400 or 7200rpm drives?

[otheos]

The 1 million dollar question, right? Wrong!

The choice is obvious:

the 7200 rpm, drives offer lower access time. They achive that in two ways: they (usually) use faster arms, and they have lower latency due to the higher rpm.

Since it is access time that gives that feel of a fast drive, a 7200 rpm will be noticeably faster than a 5400rpm drive when reading many small files (as the heads move back and forth and access time is crucial).

Large files on the other hand, if read sequentially are only affected in terms of speed by the STR (sustained transfer rate) of the drive which is dependend on the density of the platter and the rpm. So a 40GB/platter drive will be faster at 7200rpm than 5400 but the difference will not be as noticeable as with small files.

Moral: use a low access drive for OS/Programs (SCSI is ideal, even older SCSI drives outperform new IDE 7200rpm drives wiht 2x the STR due to their low access times). You can use a cheap 5400rpm for static data (MP3 etc) or a faster 7200rpm for data you handle (video/sound/image editing).

It all depends on your budget, but since the HD is the slowest subsystem in your computer (and will always be) ,always upgrade it first and buy the fastest you can otherwise it will hog the rest of your hardware.

[jadison]

Great advice, Otheos. I've noticed 10,000RPM drives have been out for a while. When do u think the prices for these drives will become reasonable? Is there much of a performance increase over the 7200RPM drives to validate an upgrade?

-=jd=-

[otheos]

10K drives are SCSI only. Comparing 7200IDE with 10K SCSI is copletely unfair, but the perfornance boost from 7200SCSI to 10KSCSI is amazing!

10K IDE drives will only come after SCSI has moved to the 15K for mainstream (now only very high end drives are there), and when the price is right. At 10K it's expensive to make heads/arms to read a 40+ GB/Platter drive, so it'll take time.

[jadison]

I apologize...:eek: Dumb mistake, yes i realize now that I read they were SCSI.

I've seen some 10K RPM SCSI go for $99 HERE (http://store.yahoo.com/a4technology/91scsihardri.html)

But, ur right, I'll stick w/IDE, until the need arises for SCSI! THat'll be when I get my network @ home up and running!

thx

-=jd=-

[hawkeye177]

My friend is so lucky. He has those 10K drives in his computer.... :mad: :mad:

[Hickman]

I've had better luck with 5400 rpm drives not getting bad sectors for some reason?

[otheos]

I've had better luck with 5400 rpm drives not getting bad sectors for some reason?

I can acertain you, nothing to do with the drives. 7200rpm drives are as likely to fail as 5400rpm (with the exception of IBM's embarassing GXP75 series) and same goes for bad clusters.

[NDC]

Speaking of 5400 vs. 7200, I just switched the hard disk out of my wife's Celeron @ 945Mhz from a 5,400 Maxtor drive to a Western Digital 7,200 drive and the difference is like night and day! Win2k loads so much faster! For anyone who has doubts that they may not see a differece from 5,400 to 7,200, think again! ;)

[jadison]

I just checked in @ an online merchant, and I just happenned to find a 15,000RPM SCSI drive for sale. I'll see if I can't find the link...but 15K of raw speed? :cool:

-=jd=-

[NDC]

Here are some 15K SCSI drives...

http://www.pricewatch.com/1/26/3045-1.htm

[otheos]

I'll give you some figures to make us all drool:

The seagate x15-36LP has
Seek time: 3.6ms
Max STR: 67MB/s

Not is it worth $380 for 18GB? Well forget the 67MB/s, just think of the 3.6ms seek time!....... it is!

[Axel]

depends on funding and how much space you need -

If you have unlimited funding - and don't need much more than 10 GB for your OS and core executibles - then the way to go is a solid state drive. No moving parts - Flash RAM - no spindle failures - speed to rival the SCSI drives - very low power draw.

now THAT's the way to go......

if you want to get really nuts - solid state drives in a RAID array - either mirroring or RAID 5 in hot swap drive bays. At that point, the choke point becomes the system I/O bus and not drive seek time.

They no longer quote prices for solid state drives that I've been able to find, in part because memory prices are in such flux. The sites typically indicate call for a current price. I do know that about 2 years ago, a 3 GB solid state drive cost around $16,000.00 Memory and costs have come down and drives larger than 12GB are out there.

Anyone who does find a price listing, please post it for us - I'd love to see it.

[Axel]

heres the kind of thing I'm talking about
http://www.bitmicro.com/products_fc.html

select products and look at the e-disk solid state tech - check out the seek times

[Axel]

here's an article for the curious -

http://www.itworld.com/Net/2784/ITW2947/

here is another one -

http://www.hewlett-packard.ca/products/static/a5294a/features-en.html

and this one appears to have a ton of good links on it
http://www.storagesearch.com/ssd.html

did find some prices - and these are somewhat current -

1 GB quantum solid state drive - $19,200
2.1 GB compaq solid state drive - between $32,000.00 and $35,000.00

so - like I said - if money is no object.....

[Axel]

And for those who have way too much time on their hands and way too much money - here's the current ultimate storage device for a bit over $212,000.00

http://www.superssd.com/products/ramsan-520/

now this IS something worth putting a little neon lighting around just for grins.

[NDC]

All I can say is... "OMG"!

[speedman]

I wonder if the life of a drive is dictated by the rpm's!
Will a 5400 drive outlast a 7200 given the same circumstances?

[Axel]

yes - the RPM on the drive is a determining factor on the mean time to failure of a given drive - more rotations means more wear and tear on the bearings and bushings the drive spindles sit in. Also - it means more wear on the read write heads as they swing back and forth over the drive. The same applies to the motor driving the spindle - and there is more stress on the drive belt between the motor and spindle.

Now - with that said - most hardwear manufacturers factor in the concept of mean time until failure - Let's explain that idea - here is the concept - you take 100 drives and you run them constantly and keep records on how many hours they run before they fail. You "simulate" the typical read-writes from and to the drive that typical use exerts. You add all the hours together and divide by 100 - you get an "average" time to failure. Typically - if the manufacturer identifies that the first drive or two that failed was because of a factory defect, it is taken out of the equation - so it would be 98 drives or so - This is a gross oversimplification - but you get the idea.

I'm not sure what the "average" is - most drives must survive the equivilant of 3 to 4 years of "average" use - I.E. the expected life-time of a computer for Joe Q. User. Another reason most computers have a 3 year warranty....

The concept called entropy dictates that all complex systems eventually break down. The more force or energy within a system - typically - the faster it breaks down. So - it follows that the more RPM, the sooner the drive will fail.

Manufacturers include better components in faster hard drives so that the drives have about the same mean time to failure as the 5400 RPM drives - otherwise manufacturers wouldn't touch the drives. They'd be paying out too much money on warranties to make a profit.

So back to media disk drives vs. solid state drives. - If you have a hard drive that thousands of users depend upon to do their jobs ( say, for example, the central computer for Federal Express - if there were such a thing ) you'd want to shell out the money for solid state drives on that server to have an incredibly high mean time to failure so your people could keep working. You schedule maintenance work on Sundays and holidays to take such a system down so the fewest amount of people have to stop their work while the system is off-line. After all - , if you are talking about 80,000 people making union wages, think what it will cost you to have them standing around for an hour waiting for the system to come back up - say $15/hr - that's 1,200,000 dollars wasted every hour until you get it fixed when it would have only cost you $212,000 one time to make sure that you never blew money on unproductive hours.

This example is why you have redundant server farms running SCSI hard drive arrays set to some RAID disk level so that when ( not if ) something breaks, your people keep working because they are your most expensive single cost every day in most businesses.

Access speed is also a large factor - servers have memory in the giga-byte range and from one to 32 processors in one box. ( IBM and SUN are working on putting 64 processors on some mainframe systems, but are scaling back because users want more servers with fewer processors for fail-over and to reduce some licensing costs ) Some database software is actually licensed based on processor count - not users and not servers - it gets more complicated from there -

Again - solid state drives have a higher access speed than the through-put of most system I/O buses. They are not the bottle-neck.

[Welsh Wizard]

One thing I have noticed with 5400's and 7200's is that they are best not mixed on the same IDE channel, as the 5400 seem to pull the 7200 down, if you want speed I tend to go for a SCSI drive with and adaptec card ( favoured choice if money available is the Seagate Cheeter, 15000rpm)

WW

[Herb]

Heck, since I read all this, I know why you're xtreme

This ramsan-520 would make up a nice MP3 repository for our network though :D:D:D

[Axel]

now that's what I call an audio-phile - someone willing to spend $200,000+ on their MP3 player - :D

[Herb]

Spending ? Not me, can't afford that, but I'd like to have it (and the other 1200 on our LAN too) :D Rock'n Roll all over the place, wow :):D

[Axel]

added benefit is no media - so no matter how many speed-bumps you cross- your player won't skip.....