Slashdot Mirror
Western Digital Pulling Out Of SCSI HD Business
leiz writes "This article on Yahoo says Western Digital is pulling out of the enterprise hard drive business. This means they will no longer produce SCSI hard drives and Western Digital will be instead concentrating on the IDE and software business. What does this mean for the SCSI market? With 7200 rpm UltraATA/66 hard drives catching up in performance to SCSI HD, products such as the Fastrak RAID 0, 1, 0+1 card, and the cheap cost affectiveness of IDE/ATA, is SCSI no longer necessary for desktops / workstations / small servers?"
Not a chance. SCSI will always be superior.
will be instead concentrating on the IDE and software business
I didn't realize they had any software.
you are so lame, you are so lame
They made rather failure prone drives IMHO. So does Maxtor. On the other hand, nothing beats an IBM drive.
I've been considering SCSI hard drives (mostly for the reason SCSI controller cards can support upto 16 devices) but have been put off by the prices. The increased performance is a plus, but at these prices? Where does that leave me, with 4 IDE devices and no room for expandability (well, none that I know of)?
... I was looking at IBM anyway.
Re: WD leaving
Now that is flamebait if I ever read it.
SCSI is vastly superior to IDE for any application, single user or not!
Have you ever used a SCSI system? Have you ever experienced how much faster, smoother it is? I use 10,000rpm U2W drives on my machine here at home. I went SCSI in July of '99 and have never looked back. The performance difference is incredible!
SCSI is better/faster than IDE, but it's *way* to expensive. If there was a 10-15% price difference, it would be worth the extra performance. But unfortunately, the price difference is much bigger (plus you have to spring for a SCSI card), and it isn't worth it.
For a server or high-performance workstation where you need to get every ounce of performance out of it, go SCSI. Anything else, go IDE.
being the purist that i am, scsi still holds more appeal for me.
putting scanners on a parallel port? bah!
putting CDRs on parallel? bah!
i like having one damn bus for everything. a nice clean interface. no stupid driver problems.
scsi makes sense.
and i'm pronouncing that "sexy."
I don't know about the lowering importance of SCSI drives for small workstations/servers (I think I'd still want SCSI on mine) -- but I do know that Western Digital makes some pretty crummy IDE HDs, so it's probably good that they're leaving.
Why is the lesser technology winning here? Why didn't people always use scsi controllers on the mobo instead of IDE controllers? Is there something about a SCSI chipset that is more cost prohibitive than IDE, or should widespread adoption bring the price down?
The current implementation of SCSI is 160 Mb/Sec. SCSI is a multitasking i/o subsystem (simultaneous read/write), IDE is not. SCSI typically impedes CPU performance by 3%, while IDE typically impedes CPU performance by ~25%. For Starters.
For small office servers, ide is just fine, infact, i tried scsi, and the the box just crashed all the time (ok, like once every 20 days or so, but thats still bad), but for a serious server, like web/mail/news/MAJOR file server, i would perfer to use SCSI, mainly because of all the neat things i could do with it, like 7 drives on one controler, or if its sub-addressed, some huge number (to tired, and to happy to see more snow to think of it) of drives.
-LW
IDE is fine for a small single user machine, however SCSI is still a much better choice for servers and network file systems. When the computer accesses data on the hard drive through the IDE bus the bus is siezed and no other request can be made. SCSI on the other hand allows mutiple outstanding requests. Data is sent from the drive when the data is available. This is one of the reasons you don't see many (if any) IDE based RAID systems. This is also why servers use SCSI instead of IDE. Although IDE is much cheaper, SCSI is a better soultion for systems services many unrelated requests for data from mutiple drives.
On a Celery 400, copying to or from an OLD SCSI drive used less than 1/4 of the CPU that copying to or from a modern 8.4G (Quantum Fireball) IDE drive did.
Now, if only they didn't cost nearly double!
Got time? Spend some of it coding or testing
Come see Mr T teach CmdrTaco a lesson!
I pity the fool who don't like mr T!
Mr T vs Slashdot
Go Read it suckas!
Christ, I'm getting so incredibly sick of EIDE. First: A fast 7200 RPM drive will deliver no more than about 8-10 MB/s. Because of the brain-damaged nature of EIDE allowing only one device to talk at a time, anything beyond EIDE/16 has been useless dickwaving. Second: It's a creeping evil. Plextor has recently released an EIDE CD-R. My local Microcenter has completely stopped selling SCSI drives. They only stock Maxtor drives, as well. SCSI is no more expensive to produce then EIDE. IBM is at least good enough to not shaft people too for buying SCSI, and their UltraStar drives are the finest hard drives that can be had for love or money. But people will continue to buy crap, driving quality out of the market. Before too long, you won't be able to buy quality at all, or at least at anything approching a reasonable price.
The blame goes on greedy companies like Adaptec who have overpriced their SCSI controllers for years. The economics of SCSI is not too good when you factor in the controller and the premium paid for SCSI hardware. It's funny when you can find excellent processors and motherboard combos for $150 with IDE interface built in, while Adaptec will sell you a controller for $300. Companies like NCR aka Symbios aka LSILogic with their more reasonably priced controllers struggle to stay afloat nonetheless.
If SCSI dies, then it is another trophy for Bill Gates and the Wintel snot box monopoly.
Look at their stock over the last 3 years (win32 only)
Mark Duell
You can get a 36 GB IDE drive from most drive companies now and rumor has it IBM has 50 and 70 GB IDE drives now (I don't know if that's really true, I'll have to check it out.) How much hard disk storage do you actually need? Seems to me you could slap 2 or 3 50+ GB drives in your system and be set for the foreseeable future (The next 5 to 10 years) unless you're storing a hell of a lot of live video goat porn or something.
I'm trying to teach myself to set people on fire with my mind... Is it hot in here?
Western Digital
:); }return(0);}
Caviar (tm) 280
AT Compatible Intelligent Drive
WDAC280 Drive Parameters: 980 cyl - 10 heads - 17 spt - 85.3MB
MDL: WSAC280-32M
P/N: 99-004071-02
CCC: A3 24 NOV 91
12V =-=-= 0.4 A
5V =-=-= 0.2 A
Made In Singapore
E 101559 LR 68850
DCM: BEABTF
Well they still make -IDE- drives
#include <signal.h> \ #include <stdlib.h> \ int main(void){signal(ABRT,SIGIGN);while(1){abort(-1)
OFTC: By the community, for the community
SCSI is targetted at high-end users in need of reliable drives. WD is cheap, but has a horrible reputation for failure rates. Few of the people buying SCSI would also want to buy WD.
For what I can get 18 GB of SCSI and a decent SCSI card for, I can buy _4_ 36 GB IDE drives AND a system to dedicate as a file server. Even if setting my system up as SCSI were to double my performance, I'd have a hard time justifying the extra expense.
I'm trying to teach myself to set people on fire with my mind... Is it hot in here?
Someone moderate this crap down. I don't want to see this as the first thing under an article on a hard-drive company, f'Chrissakes.
SCSI used to be more expensive because IDE hard disks usually had lower RPM and cache than their scsi counterparts.
However, now there are a lot of IDE disks which have similar stats as the scsi hard disks. However, scsi is still much more expensive than IDE. Does anyone know why?
Tim
Hey, send your proofs to intel.. .they'll hire you in a heartbeat to do their next set of proc benchmarks. (SCSI, dedicated proc on the controller, 160MB/s. IDE, all processing done on the host proc, 66MB/s BURST)
I spent seven years at Western Digital, and I watched a chip company with an amazing amount of IP (basically everything except processor and memory) scuttle the chip business on the alter of the almighty hard drive. WD has seeded many successful companies in SoCal (Broadcom, QLogic, Emulex, Silicon Systems, Adaptec, JNI, more...) by letting go their talented engineers when management had failed yet again to follow in a market where they should have been leading. I'm sorry to have seen it happen yet again. Sounds like sour grapes, but it's really not; I'm just not surprised at this latest news. It's just another round of layoffs at WD, which isn't really anything new there. To any of the old guard still there -- you're a hardened lot, and I wish you luck in yet ANOTHER new direction set by the company. As for me, let's just say that in retrospect, WD was a great place to be from.
You don't use science to show that you're right, you use science to become right.
I am the proud owner of a defective western digital 7200 rpm UW scsi AV drive. (it makes this really neat ka-chunka chunka chunka chunka sound when it boots up :)
it seems the "AV" drives are a problem for them.
I called Western Digital and asked why the heck i hadn't received my replacement after waiting 3 weeks with an RMA # and the tech tells me they make the AV version of the drives "by hand" and they don't expect to get any more in until sometime in february. i should call them tomorrow eh?
--freq
"Tension is the great integrity" -- R. Buckminster Fuller
I've been jacking around with hardware since '85, and the only two hard drives that ever died on me were both made by Western Digital (a 1gig ide and a 3gig scsi). Since then I stick with Seagate and IBM drives for my machines. I won't miss Western Digital's abscence from my favorite periphrial interface a bit.
linux and intel smoke sun/sgi/hp in the enterprise market, its just their FUD that keeps people believing them...
Then where, pray tell, should it go? We have no other public forum in which we may air our greivences, and until there is a such a place, I have no choice but to make my voice heard here.
If there was a permanent forum where all such postings could be discussed, then perhaps we would use it (and the sid comments don't count: they are neither permanent nor well known), but there is no such thing. Maybe that's something else that Slashdot needs to fix.
~~~~~~~~~
auntfloyd
First off, I'd suggest some research on your part. SCSI has many advantages over IDE, EIDE and any implementation of UDMA/** with the IDE form factor, depending on what your trying to use it for. Will you see a massive advantage on a small home user system, not really, unless maybe your a Quake3, Unreal, or other massive game player.
However, attempt running a server of any kind with roughly 200 users pulling about 5 apps a piece using IDE/UDMA drives. What you will get is really bad load times. SCSI has the ability to read/write at the same time, not to mention it can read/write to all the drives on the chain at the same time Vs IDE's ability to read or write on one drive per BUS. Hence why you get better performance from having Ur HD and CD-Rom on sperate IDE buses.
However, I think that WD is possibly pulling out of the SCSI HD market for reason like, high pricing causeing low volume sales, the explosion of home networking, or they are just tired of making SCSI drives for other venders. It's not often you find a WD labeled HD. But the recent explosion in home networks is what I'd bet some of my money on. Now with the average of 2/3 pc's per house, that 2/3 times the ammount of HD's being used and as we all know, most home systems containe IDE HD's.
Each of the two formats IDE - SCSi has it's advantages and dis-advantages. It just all depends on what you want to use it for, and how much you have in you wallet. Still though, it's sad to see a company give up on a product after so long.
Trying to be different, just like everyone else.
Have you done any testing yourself or are you just regurgitating some rumor that your friend's brother's dog's pet parakeet squawked one afternoon? How about listing your sources of these benchmarks so we actually have something to refer to.
High end systems will be using SCSI for a bit.
1. ATA-66 can't touch 160 mb SCSI, even single drive to single drive, with serial access.
2. ATA performs very poorly when multiple reads and writes queue up.
3. SCSI handles large numbers (>4) drives much better, ATA sees problems with just 2 drives per channel.
Maybe ATA-262 will have a shot.
-Peter
When I was in the market for a CD burner, everybody was saying "you need SCSI, it's faster". Well, 4x is 4x. I bought an internal IDE burner and have yet to make a coaster. I figured they wouldn't sell them if they were that bad (although to be on the safe side, I went retail box rather than OEM).
So, am I missing something? Are there situations where 4x is really not 4x?
As for hard drives, I understand that SCSI drives are faster, but drive performance really only slows you down when you are thrashing, and if you are thrashing you should get more RAM not a faster HD. I suppose this makes sense on servers where there is too much data for everything to be in RAM, but I don't see a need for SCSI in my workstation PC.
For all intensive purposes, "whom" is no longer a word. That begs the question, "who cares"?
I know everyone keeps telling me about how good the new ATA66 is, but I'm a little skeptical -- because everyone I know that runs it swears it is as good as SCSI. I've run both IDE & SCSI over the years and I know that SCSI runs circles around IDE any day of the week. No CPU getting sucked dry, no worries about speed, and no worries about how many drives to hook up (if you get worried about hooking up over 15 drives, you have more worries than I'd like to hear about ;-)
.sig is a Rolls Royce.
I find the worst thing about IDE is the limiting of your drives to 2 per IDE interface. This is about the worst hell for me sometimes. I mean, a CD-Rom, a CD-Burner, two hard drives (because you know you have to much crap on the 1st and you'll ned to get a second). And when you realize both those drives are full and you've burnt everything onto disk with that slow assed IDE CD-Burner that you could dream about, a 5 IDE interface would be real nice. (of course, you could probably use an old IDE interface, but are they worth it?)
Maybe with a big gun like WD totally dedicated to IDE technology, IDE will get an even better boost than the ATA66? Who knows, maybe they have some tricks up their sleeves...
[Crimson]Chain
My other
WD always made bad SCSI drives, anyways. Their old ones, like ones were slow, noisy, and were heat problems. Their new ones were always considerably slow compared to their competition. WD was known for their IDE drives, but that reputation isn't to strong anymore.
I've got 4 SCSI one drives, a SCSI-2 Seagate Barracuda, a UW SCSI IBM (9zx), Yamaha CDRW.. my brother has two Seagate 4gb, and two Yamahas, and my mother has a SCSI caddy cdrom. And of course there's lots of IDEs laying around (hdds, cdroms, CDR, DVD), and a few MFMs, plus a prop. 1x sony cdrom.
Problems with SCSI:
1. Barracuda came defective. Same with Yamaha. Both from a really bad reseller who gave me a bad controller (defective), and claimed Seagate's tech. was lying and the drives really were out of production. Took 3 months to clear up.
2. IBM over heated, and eventually died months afterwards. IBM replaced within a week, the data was recoverable. The drive was 1st generation 10k rpm, and a pre-release w/ updated rom.
3. For some reason my brother's Yamaha CDR102 wont write onto newer cd media. Seems to be a cd-design change, as old media works fine.
4. Always requiring innovative tweaks to keep cool. SCSI caddy cdrom (6Plex) overheated and was replaced for free, years ago.
5. Pain to get UNIXes to install with the Yamaha. They'll boot off cd, and then say there's no CDROM to install from.
On non-SCSI problems:
1. Connar 1.4gb drive had bad blocks, years later stopped functioning.
2. 8x cdrom 'kinda' works.
3. 10/12x cdrom sticks.
So, there are more difficulties on SCSI, but cabling and installation is easier. Cooling is just a pain. Performance, though, is high. My Barracuda 2lp still outperforms IDEs in cpu/speed. The IBM is so fast I feel bad having it... Still, its amazing seeing 10% cpu used, max, when 100% is used on IDE, like the few DMA/33s I have.
SCSI for home = waste
SCSI for workstations = ok-good (mostly useful if the CAD is CPU oriented)
SCSI for servers = good-great (depends on what the server is doing, load, etc).
Just an anon that's really bored....
I think its safe to say that Firewire is pretty close to SCSI in terms of performance. Which to say, blows away USB and IDE. Fibrechannel on the other hand, is faster - but I suspect that it will be the "future SCSI". Always faster, but always more expensive.
Given that Firewire is intended for more light-weight devices (digital cameras for example). I suspect that the chipsets are pretty cheap. And assuming licensing costs are reasonable, then the actual drives should be pretty competitive.
Just snooping around the net, I found a 28MB external Firewire drive for $500. Not a bad price for a new technology. (keep in mind there usually a premium for early adopters)
Tom
I've found the IDE performance problems go away entirely with the correct bus-mastering settings. You still can't get decent performance with two drives on a single channel, but my new Abit motherboard came with four IDE channels. So you can have four IDE hard drives without losing performance.
Here are some real benchmarks to back this up:
[root@olympus /root]# /sbin/hdparm -c0 -d0 -k1 -m0 -W0 /dev/hda /root]# /sbin/hdparm -t /dev/hda
/dev/hda: /root]# /sbin/hdparm -c1 -d1 -k1 -m16 -W1 /dev/hda /root]# /sbin/hdparm -t /dev/hda
/dev/hda:
<snip>
[root@olympus
Timing buffered disk reads: 64 MB in 28.40
seconds = 2.25 MB/sec
[root@olympus
<snip>
[root@olympus
Timing buffered disk reads: 64 MB in 4.86
seconds =13.17 MB/sec
That's with a 7200 Seagate drive. The first benchmark, giving a whopping 2.25MB/sec, was with all the IDE options in sucky mode. This is the way older IDE controllers work, and in large part responsible for IDE's bad name. The second benchmark shows that it can have good performance. It's CPU performance wasn't as good as SCSI's (17% out of 200%; dual-processor box) but wasn't as bad as many have said.
Why moderate it down? This person has a very valid point that will not find itself in a headline. Realy, Rob, why not put your money where your mouth is?
The big trouble with IDE is still that they are "dumb" devices that require CPU resources to manage. On workstations doing lots of disk access I can see NOTICABLE performance degradation between similar hardware, one of which is IDE, the other of which is SCSI. The nicest thing about SCSI is the fact that the controller offloads all disk management off of the system's CPU. If you're doing power computing, this makes a big difference. Also, as someone else mentioned, IDE has real problems allowing the system to manipulate multiple drives simultaneously, a problem SCSI does not have. For some schmuck just dicking around with Netscape so they can browse the web, who cares, but for hardcore users with big machines trying to get real work done, it can make a legitimate difference.
From a server perspective, there's no question that SCSI is the best. Just TRY putting more than four IDE drives into a Linux box without tearing your hair out and threatening to take a shotgun to the thing. The only way to do it is to get some sort of additional IDE controller like the Promise controllers which are unmitigated junk. I don't even want to mention the hoops I've gone through to to get a Promise Ultra33 stable in my Linux server. What makes it worse is that I could buy the four IDE drives I put in there for about the same price as I would have been able to pick up two SCSI drives of about the same size. (It's not that SCSI is so tremendously expensive as much as it is that IDE is just dirt cheap.) More unfortunately, I needed the space and I didn't have the extra money, or I *would* have just gone with SCSI. (As it turns out, I spent so much time trying to get the IDE drives working, I probably *should* have just gone SCSI from the get-go and saved myself money in the long run from doctor's bills from high blood pressure and ulcers trying to build an IDE-based server will give me.)
I see the whole "IDE vs. SCSI" thing as yet another case of mediocrity winning the battle. It doesn't have to be great as long as it's cheap and good enough to get the public to buy it. For those of us who like quality, we just have to pay so much more. Unfortunately, unlike the software industry, there's no way to start an "Open Source/Free Hardware" movement to force the other manufacturers to start focusing higher on quality.
-=-=-=-=-
-=-=-=-=-
My mom's going to kick you in the face!
What have you been smoking? EIDE beats SCSI/Fiber? OK, and your TRS-80 can run circles around a POWER3 RS/6000 as well, right?
What a Troll!
~Any apparent grammatical or typographic errors are caused by defects in your display device.
IDE, even with all the new ATA/66 performance improvements, is still not quite up to par with SCSI. SCSI devices tend to use considerably less CPU time, since so much more is done on a hardware level.
:)
:)
SCSI has purposes other than mass storage. Because of low overhead and much wider bandwidth it can be used for just about anything... There are even ways to network computers together using SCSI (not very practical, but possible
Another thing that IDE lacks is portability - You can't hook up any external devices to your IDE bus unless you have a 3-foot IDE cable hanging off the back of your PC making it look even more ghastly... Plus, even the most advanced dual IDE controller still only allows you to hook up 4 devices simultaneously.
yet another issue, even though it applies mostly to the x86 family of computers, is the fact that you need a separate, non-sharable IRQ for every IDE channel. That, IMO, is a flaw, considering that PC's have so few precious IRQ's available...
The only serious advantage of IDE over SCSI is the cost. Although, finding a cheap IDE drive that can last for a while is rather difficult... *cough*...fujitsu...jts...*cough*
WD is most likely quitting the SCSI storage business because they can't keep up with the competition in speed and reliability... That might be a good thing, cause now they can spend a few extra R&D dollars to figure out why the damn Caviars make that deadly clanking noise and quit after only a few months
--------
--- sig moved for great justice.
Plop down $4760.00 for an Ultra[tm] 10 and you get IDE.
Ultra is a trademark of Sun Microsystems, and don't you forget it.
Oh, and don't forget My Sun . How cute.
Lurking at the bottom of the gravity well, getting old
We are simply reaping the rewards of the commercialization of a previously "good thing". Slashdot sold out (ie: made money and lost control) to Andover, and now Andover do not want to lose eyes (ie: site viewers). Once again, vision makes way for trepidation and slow degradation. So long slashdot, it was good while it lasted.
Besides which, ATAPI always seemed more CPU intensive for operations. I've not felt that after going pure SCSI, but whatever -- YMMV.
Seriously, I've never had any good experiences with WD, either IDE or SCSI, so won't shed any crocodile tears for their loss, and I really doubt that this means the end of SCSI--although it may mean the end of BAD SCSI...
Just because you're paranoid doesn't mean they aren't out to get you
With ATAPI and LBA, what we essentially could have is the equivalent of SCSI but over a cheap cable with standard TTL drivers and no fancy termination (the things that make SCSI expensive). We could get nearly the same speeds -- or at least, we could if we were willing to put a single drive on each IDE interface (which certainly isn't out of the question).
Maybe, if it's dropping out of the SCSI business, WD can be persuaded to flesh out its ATAPI command sets so as to keep customers. Wouldn't hurt to ask.
--Brett Glass
36GB IDE drives up 'ere in Canada are $500, 16GB Quantum Atlas IV SCSI drives are $600. A decent controller is another $300, but that is a ONE TIME cost and you can string a shit-load of discs on each CHANNEL of that controller. Sure you get half the space per drive, but when you are talking about a SERVER SYSTEM where speed AND reliability under high load is the goal, you definitely want to avoid IDE drives.
Ahem, you are the idiot! Of course a single drive does not do 80 or 160 on scsi. But udma won't make it go 66 either. SCSI bus gives more bandwidth for simultaneous drive access, which kills udma. Try to use those 2 ise drives a t once, eeewwww. Now consider 4 ide channels is max, take the performance hit and put 2 per channel, thats 8 devices, and 4 IRQs. Now the SCSI on the otherhand will take 15 devices on that bus. Need more? add more channels and let them share IRQs, it works, no shit! I've had 4 scsi busses in a system but thats not the limit. In all ways except price, SCSI is superior. If ide would just fucking DIE then scsi would come down in proce. FUD is what momos like you spread and it certainly doesnt help ide die. Wanna kill some old useless tech? Don't kill isa and all my ports that i need for baseline compatibility. Go kill that damn IDE!
Now that he's gone and the SCSI business is a memory, you can all expect nothing but crap to come out of this company for years to come.
All of these posters talking about EIDE (or whatever this months incarnation of the ATA spec is) being better than SCSI have no clue what they are talking about. I use my computers alot. Anytime I sit down to a system with any type of IDE drive, I can immediately feel the sluggishness set in, all while the CPU wastes cycles babysitting the rather braindead disk channel. Server or not, SCSI systems are *always* better and I will *always* continue to pay the extra quid to be at the keyboard of a system that doesn't slow me down. For me, that's not EIDE - ever.
Case in point: my shiny new Dell 600MHz system with the best Dell has to offer in EIDE technology. Many fingertip tappings waiting for the fluttering of the hard drive to settle down whilst I work. To me, that's not good technology or a good use of my time. At my earliest conveinance, I'll be swapping out the disc subsytem in favor of something with 80 pins and real bandwidth capability.
Pffft. With a 6.4 gig UDMA/66 in a very loaded (CPU + memory) Linux system at work and all SCSI-UW at home, I can tell you IDE is nowhere near catching up. You still end up with a flood of interrupts using IDE... SCSI is more "correct" from an architecture standpoint as well.
I've never really liked WD products personally, but now I can officially say I will never use another one. (-: Long live superior interfaces!
Coasters are the big deal. (I've got an IDE CD-R too.)
It's very easy to make a coaster if the data can't get to the CD-R fast enough. This used to happen all the time, before they started putting bigger buffers on the IDE burners.
Today's IDE CD-R drives are more reliable, but you can still make a coaster by stressing your system while burning. Loading a big program could do it. Playing Quake III will do it. Because SCSI multitasks better, it will be less likely to make a coaster under stress.
So you can DO things with your computer while the CD-R burns.
to here: freeslash
wow cool, score 4 hehe i have gone an created an account for myself. Mr T will make tacofool release a new version of slash!
Quit yo jibba jabberin' and come see Mr. T
On the SCSI v. IDE debate, I have to vote with the SCSI folks. As good as IDE has gotten, there are things which will just choke it, while a SCSI system will float along as if propelled by the wind.
But do we really want SCSI? Big expensive cables? Termination problems? All sorts of crazy complexity? I don't know if there is any reasonable way of comparing IEEE-1394 v. SCSI yet, but I sure think that if our friends at intel ever get on the wagon, we could probably get rid of both SCSI and IDE and be fast and free in the future.
IDE is well suited to the single tasking OS. Where will we be when the majority finally move beyond that? God help us all.
"pentium cpus are 600mhz+, while lame so called enterprise cpus from sun are only 400mhz."
This clueless statement reveals the depth of your knowlege. Megahertz is the most WORTHLESS way to rate a processor's performance. A 400mhz UltraSparc stomps the crap out of the fastest pentium available. Which is more powerful, 600 pounds of gunpowder or 400 lbs. of TNT?
Please read the white papers before you spew pure fiction.
~Any apparent grammatical or typographic errors are caused by defects in your display device.
A lot of people use this EIDE crap, thinking it's great for a server and what not, after all it works for their desktop. It's not.
While I'll admit, there are some SCSI disks which are differentiated from the IDE drives solely by their interface, the higher end SCSI disks usually do have some serious advantages.
Some of the bigger SCSI advantages are
the fact of the matter is that if you want a cheap drive, you can buy a cheap EIDE drive, or a cheap SCSI drive. if you want a *good* drive, ultra-high quality EIDE drives are virtually non-existant, leaving you with good ole' SCSI.
A lot of people have this odd notion that when two computers are PIII 600s with 256 megs RAM and 18 gigs hard drive, but one costs $500 more, that the more expensive one is automatically a ripoff. People seem to forget that sometimes the more expensive one has better components and is less likely to die and wipe out the past two weeks of work. (all you non-student types, how much did you make in the past two weeks? I'd bet a *lot* more than $500). We need less ads that say the price, and more like the great VA Linux ad with the steak dinner on one page, and the TV dinner on the other.
It's a matter of cost/performance ratio. Yes, SCSI is light on taking up CPU resources; no, SCSI isn't much faster than a good 7200 RPM ATA66 disk, and IDE is reaching into 10K RPM these days.
So, in terms of performance, is it worth the several hundred $ extra per hard disk? No. Purchasing a 25% faster CPU for several hundred extra dollars is the better long-term solution, because each time you need to add a new HD or replace an aging one you save the $$$ overhead you'd be paying for the additional SCSI drives.
No one ever said the most technologically sophisticated solution is the best solution--it's not. A 70s muscle car will generally kick the ass of a 90s sports car. The same is true with computer hardware, which is why commodity IBM clones have consistently kicked the ass of more elegant PPC boxes and even SPARCs and Alphas--sure, Alpha is the fastest thing this side of whatever the NSA's private little fab is putting out, but for the price of a smooth-as-silk Alpha or SPARC server, you could have built the ass-kickingest SMP x86 box--or two. Sure, SCSI will give you a performance boost--but you could get more processing power and more disk space with IDE 66; the processor power advantage would be nullified by the IDE pull on the CPU, but that still leaves you with more disk space.
More is better, right? So unless you want to serve pages from that pathetic 2GB SCSI disk all your life, because you can't afford to add more SCSI drives, just ride the ATA66 revolution. If you can afford mondo SCSI disk space anyway--go with IDE and get yourself some more bandwidth or throw in more RAM.
Thing is, with a computer there are always trade-offs, always several things you could get to improve performance. Few if any of us here have unlimited wallets--SCSI is dead. SCSI is the past. In the future, for the highest-end most expensive hard disks, we'll have FireWire or some other high speed standard--not SCSI. And for all other applications, IDE 66 and successors will be the way to go. The revolution's on, folks--all that stuff we've been using for 20 years is going by the wayside: x86 architecture (at least as we know it--maybe Sledgehammer and Crusoe will make it more serviceable), Microsoft operating systems, ISA slots, SCSI, and I wish my old college would hurry up and get rid of that ancient VAX, too.
"The more corrupt the state, the more numerous the laws."--Tacitus, *The Annals*
Remember, way back in the day, when Firewire was supposed to blow both SCSI and EIDE away with its throughput (480 megabytes/second, IIRC) and its cheapness?
Did it just sink in a quagmire of "industry standard" arguments? Talk about a promising technology that's biting the dust.
"Even genius needs a competent technique."--Robert Fripp
I've got to say, this doesn't suprise me one bit.
I worked for a Computer Repair shop for about 5 months now. Here's the Breakdown on the Brand Names of Drives that come in Crashed that I've seen so far.
90% Western Digital - at least 20 that I can think of offhand
8% JTS (which are out of business) - 2 of these
2% Seagate - I know of 1 that had bad sectors
As of yet I've seen no Maxtor, Fujitsu, IBM, Quantum, Samsung or any other manufacture's hard drive crash. Although I've heard alot of bad things about the Maxtor Drives and the Quantum Bigfoot's Crashing. and I know from personal Experience that some old IBM Drives, (and I'm talking 10-15 year old PS/2 Hard Drives) were crap.
We would sometimes get WD drives that came fresh out of a box, stick it in a machine, and it would be damaged. My boss Had to deal with a company for a week because They bought a WD Enterprise Drive for their mission critial Server and it crashed. It Wasn't even a year old!
If I had a choice of any drive today, Hands Down I would have to go with the IBM Drive. If I had a second choice, I would probably go with a Samsung or a fujitsu.
In Soviet Russia, Trojan exploits YOU!
I simply refuse to believe that the SCSI interface on an HD hasn't been worked down to practivally a same single ASIC as IDE has. What's more is that the boards on most HDs are modular. The underlying drive mechanism is identical. Only a different board is plugged in and screwed onto the drive to make it IDE or SCSI. Why should there be such a cost difference? They are simply gouging SCSI buyers because we tend to put toghether more expensive server type machines. So they hike prices... then sales go down... then they say, we're dropping SCSI because sales are way down. WTF?!?!
what a lame rebuttal. the fact is, pentiums do in fact blow away any sun cpu on the market. both in integet and fp and spec.
a quad xeon p3 600 smokes a 30cpu sun E6500 with their latest 400mhz cpus. sun cpu performance is pathetic...and their hardware is so shoddy and unreliable its a wonder anyone buys there equipment.
facts are, linux and intel dominate the enterprise computing arena.
EIDE will eventually hit limits as even desktop computers become more demanding. As that time arrives SCSI will take over. On the server front, fibre channel is looking like the future over SCSI. It may be even more expensive but it is faster, has a crazy 10 km or so distance limit, supports more devices on one loop, and supports multiple HBA's connected to one set of devices. This allows multiple systems to talk directly to the storage rather than through a network to a computer that talks to the storage. SANs are going to really need fibre channel.
SCSI may seem to be "too much" for the average user but in that as the old MTV logo used to say..."Too much is never enough!" This held true for music television and it holds true for computers. I remember getting time on a 386 SX 25 Mhz with 4MB RAM and 80 MB hard disk. This was a $10,000+ system at the time. Now it's a paper weight for all but a few geeks (like me) who love to find uses for old hardware. I have a few IDE paper weights...I will have a few more before they are done.
-- soldack
Hey Transmeta! Let's have SCSI ports standard on the Crusoe mothermoards for desktop machines when you come out with them! And no IDE ports! Let's spurn Wintel and make this stuff cheap through mass sales/production!
It's flamebait because he posted first. Many moderators would just like to get rid of their points ASAP, so they see the first post being only one line long, posted by an AC, and they think "Troll, Offtopic, Redundant, Overrated. Who cares, I can blow a point."
:(
Almost guaranteed the mod didn't read the context of that post
-- posted by a responsible moderator, which seems to be why I get mod access more than I'd like to =)
If you're referring to your personal machine, perhaps you're right, assuming you love to do system maintenance. In a corporation, the cost factor definately swings towards SCSI and RAID. While they may seem quite expensive, the fact of the matter is that downtime costs are usually in the thousands/hour for small companies and go up from there. (put 200 people with an average salary of $70k out of work for an hour, and it just cost you $7000.) Cheap hardware starts looking really expensive when you realize that.
You can argue that your personal system downtime costs nothing, but that assumes that a) you don't value your time or you like spending it rebuilding systems or b) that you don't do anything worthwhile on your computer anyway. If these are true, then buy the cheapest thing you can find. Otherwise, I suggest looking at the MTBFs and realizing that about half the drives fail before that time.
--
"One World, one Web, one Program" - Microsoft promotional ad
"Remember when the U.S. had a drug problem, and then we declared a War On Drugs, and now you can't buy drugs anymore?"
I laugh at da foo who can't beat a taco!
So responsible you seemed to have forgotten the big sign on the Moderator Guidelines page... I dunno, something about anonymity?
Open Source. Closed Minds. We are Slashdot.
That is just plain immpossible. Comparing Apples to Apples (non-raid to non-raid, raid to raid) Fibre beats SCSI and blows away IDE. Totally. Completely. Always. The card I work on pushes over 30,000 I/Os per second without a problem. 300 MB/s? No problem. No IDE system is doing that.
-- soldack
Ever hear of SpecInt?
~Any apparent grammatical or typographic errors are caused by defects in your display device.
This belongs in quickies ;)
/. -- omg, Mr T. working for (I won't say it to ruin the fun ;) is hilarious =)
I never laughed so hard at a link from
Most Linux setups I've seen default to using at least a couple of the "sucky" IDE controller settings. The "-d1" setting with hdparm is crucial, in particular, as it turns on DMA, which hacks a huge chunk out of your CPU usage, makes things nice for the scheduler, and increases transfer rates dramatically. With DMA off on my system MP3s skip whenever the hard drive thrashes too much; with DMA on I can't make an MP3 skip with any level of hard drive activity (and believe me I tried).
One last flag you might want to try: -X34 will make sure the drive is set to DMA mode 2 transfers, and on new drives -X66 will select Ultra DMA transfers. DMA->UDMA isn't nearly as big a leap as PIO->DMA, but it's sizable.
I wish more people knew about hdparm - it's a single command you can run as root that can double the performance of your system under some circumstances. I think new kernels are getting more aggressive about enabling good IDE settings themselves, but there are still too many systems out there where the default settings needlessly give both Linux and IDE a bad name.
IDE can't switch between master and slave fast enough to allow the greatest performance increases when striping with RAID and swap. It has silly limits like 2 drives per channel. It does'nt support command re-ordering in hardware to allow the heads to move less during many access'. It is not multi-threaded... blah blah blah. If this is true, I have lost respect for WD. My Caviar 340Mb is still going strong and I loved their build quality. I know IDE is getting really fast now, and it's cheap, but for the server with really disk heavy applications, transfer rate it not the be all and end all. Neither is how fast the heads can move. SCSI is far better for server stuff. Damnit!
War crimes, torture, lies, illegal spying... Would someone give Bush a blowjob, already, so he can be impeached?
I've been using SCSI for about 5 years now, when I bought all these external hard drives for my Macintosh. Over time, I eventually moved to a PC and could take these drives with me. Since then, I've been using SCSI religously.
I get better performance out of the drives, and they're 95% quieter than any of the IDE drives I've purchased.
As a digital audio fanatic, I swear by these Seagate 50gig SCSI drives and would shoot myself if I ever had to use IDE for what I do.
Failure isn't falling. Failure is staying down.
Ultra160 specs
Lol.
/. reader, logged in, you'll get mod access. It's a given fact.
I get (key word: get) mod access, but it's not like I go telling people "okay, I've got mod access, what post do you want me to mark up?" Me telling you that I get mod access is like me saying that I click on the articles to read them: if you're a regular
I never divulge which messages I mark up or down, that's my business and mine alone (and will get judged in M2). I don't have mod access today, and as such I don't have mod access for this discussion -- therefore what exactly did I do to break the mod agreement?
Please tell me, I'd love to know.
I personally see Western Digital staying afloat because they have contracts with big name computer companies and mindless computer buyers with no preferences on hard drives inadvertantly buy them. Kinda like ATI staying afloat on the macintosh side. ATI video is in every freakin macintosh coming off the factory line today. They suck performance wise on the PC side, and any performance enthusiast will stay away. See what I mean? I agree with the general posts of WD - they suck.
"a quad xeon p3 600 smokes a 30cpu sun E6500 with their latest 400mhz cpus. sun cpu performance is
pathetic."
This is an obvious troll. I give up.
~Any apparent grammatical or typographic errors are caused by defects in your display device.
Personally, I'm using SCSI as a replacement for 'traditional' ATAPI devices in my system. My CD-writer, DVD and CD-ROM are all SCSI, freeing up my IDE channels to handle hard drives. The only thing stopping me from all SCSI is the prohibitve cost. $600 cost for a 18GB WD 10K is crazy. $600 gets me a lot of IDE storage space.
Personally, I don't view this as that much of a loss to the SCSI world. Western Digital has been on my bad list for quite some time. The past several hard drives that I have had fail on me have all been Western Digital,and usually just past the warrantee (dohh!). Lately I've been buying mostly Maxtor and have had pretty good luck with them. I've also had good luck with Quantum (but I've heard lots of bad things about their BigFoot line -- never bought one myself though). I've been mostly buying IDE drives lately, but when I buy SCSI drives I generally either get Quantum or Seagate (although I am not a big fan of Seagate's IDE drives, their SCSI drives seem pretty good). I've heard good things about IBM's SCSI drives, but haven't tried any of their newer ones.
All in all, there are enough other choices that I don't think Western Digital will be missed in the SCSI world.
SCSI is expensive, and a card still $200 or so, but you're exaggerating.
Open Source. Closed Minds. We are Slashdot.
IDE suits me just fine. SCSI is too expensive for what I want to do. My box doesn't function as the world's largest file, database, web or even DNS server (nor does it come close), so I'll stick with the bext combination of cheap and reliable I can find.
To the earlier poster who said "if people keep buying crap, they'll drive quality out of the market," explain to me how it is that BMWs routinely out-sell your basic rice-burner if crap drives quality out of the market. :)
You know SCSI 160 is totally useless unless you have a drive that can saturate that channel. So far, there's no single drive that can, so you look to RAID options. Right now, there's ATA-to-SCSI adapters such as the Medea setup that my friend is using. It lets you hook up ATA drives to an Ultra2Wide SCSI card, so you're getting an equivalent of a SCSI RAID using much cheaper ATA drives. Because of stuff like this, SCSI drives are losing more of its edge over ATA drives. Of course, you'll still need SCSI for those dozen-removable-drive servers and for external stuff, but at least now there's cheaper alternatives to some of the stuff SCSI was good for.
please, do not immediately say I don't know what I'm talking about, I do (somewhat, no real life scsi experience, but i do have a 3rd generation seagate barracuda in the garage)
ata hard drives have historically been slower than scsi hard drives. But recently, they have been catching up. ATA/IDE drives now use busmastering and dma, making the cpu utilization less. The speed of the interface has also been increasing, up to 66 mb/s, beating ultrawide scsi 2 which was state of the art about two years ago before lvd scsi arrived. Of course, this is just the interface itself, the actually speed still depend on the hard drive itself. A few years back, IDE hard drives were running at 5200/5400 rpm and had 128k/256k buffers. Currently, IDE hard drives run up to 7200 rpm and has 2 megs of buffers and a 10 gb/in^2 areal density, whereas SCSI hard drives have up to 10000 rpm (giving them lower latency and stuff) and a 4/8 meg buffer (mostly on AV optimized models) but only a 3.3 gb/in^2 areal density. Anyway, the point of all that was to say IDE hard drives are catching up.
Back to the interface:
scsi can go up to 45 devices per card (3 channel UW card, 15 devices per channel) or more realisticly - 15 devices per card, on a single channel UW scsi card - totally beats IDE's 4 devices per card/motherboard limitaion. but what about those ABIT motherboards with 4 ide channels or the ability to add on a PCI ide card? you can argue this will use up all the irqs but PCI irqs can be shared. Then there's USB/Firewire, which allows up to 127 and 63 devices respectively. USB ports are standard on all PCs since around 97 and they can be used for low speed pheripherals such scanners or zip drives. Firewire is available on various compaq/other name brand pcs, as well as some Macintoshes and can be added onto any pc through a firewire card (adaptec makes them AFAIK and they are commercially available) and firewire has the bandwidth to support high speed hard drives and other pheripherals. With all this, shouldn't scsi be obselete by now? (Yes, I know scsi hard drives are still the fastest, adn I understand SCSI is necessary for hardware raid 5, and SCSI hard drives have a higher MTBF - all of this is critial to mid/high end servers) but what about the low end server / workstation market? (where price may be an issue, btw, I'm ruling out the home pc market because the _ordinary_ user doesn't have 5 hard drives and a scanner, and a cd burner, and etc, etc)
which brings me to my next point:
scsi is incredibly expensive. yes, there are cheap (7200 rpm) scsi hard drives out there (still $200+ though) and there are cheap scsi controllers (tekram - $170 aint bad) but 7200 rpm scsi hard drives dont give the performance advantage to justify the additional price (10000 rpm hard drives start at $350) and the tekram scsi cards (they provide linux drivers, cool eh) are only up to 80 mb/s, not 160 mb/s (the new adaptec cards - which are REALLY expensive - the scsi card alone can buy 20gb - 30gb worth of IDE hard drives) so looking at the cost effectiveness of the scsi hard drives - they are not really worth it for low end servers / workstations now are they? (the cost is worth it for high end servers, especially if they run anything critial where maybe a business is depending on the speed and reliability)
just a little note: by "workstations" i'm also including people's high end machines and those ultimate gaming rigs...
now, back to slacking off, for I've got senioritis maximus =)
_______________________________________________
There is no statute of limitation on stupidity.
You know you could find yourself in jail for selling what ever it is you're on?
Open Source. Closed Minds. We are Slashdot.
Face it, the IDE design is ancient and is inadequate for today's uses. I mean its fine when you are plain ol' Joe Bob who just checks his email and does word processing. But when you want to *add* something to the computer and do some serious stuff, like a geek will, you're running into problems.
Not only does IDE have bad command queuing, it doesn't even do sync transfers. The most debated issue is the CPU usage and the transfer rate: IDE relies on the CPU more than it should because the controller is too simple and therefore braindead. You can overclock the IDE controller, but what always happens is the drive is too crappy to even handle the higher speed, but you always get the same read performance. IDE always sends date from devices back to the host controller at the same original spec speed, whereas write-to-device can vary due to oc'ing the controller.
Ok, now to the point:
the engineers (I'm sure they've been TOLD to do this) keep trying to make IDE "better" by keeping this backwards compatiblity junk and at the same time trying to squeeze a wider data bandwidth for the devices. Think about ATA/66, you need those special 80 wire/40 pin cables because if you used a regular 40 pin/wire cable the signal to noise ratio will be so bad that you get tons of CRC errors. The additional wiring are for the extra shielding in order to keep the SNR well enough to avoid CRC problems. What about adding additional devices for more storage space and removable like what most of us geeks do? Okay, they draw up these brilliant schemes of secondary, tertiary, and quarternary controllers which are essentially the same in controller design as the "primary" except on a different IRQ and port. Wow, cool, now I can hook up 8 IDE devices!
Ok, but I want to add some stuff like: a PCI soundcard (2 IRQs... 1 for ISA/DOS emu, and 1 for actual PCI), add NIC (there goes another IRQ), add DVD decoder card (1 IRQ). Hmmm... wait a minute, isn't IDE 0-3 using IRQ 10,11,14,15 already? So didn't that left me with IRQ 9 for video? Ok, suppose I _DON'T_ even have a NVidia based video card (which has problems sharing IRQs), and try to share IRQ 9 through "PCI steering" with the USB, also; that only gets me 2 devices working. I still have to disable the serial port(s), and the parallel port to get more of this working. It is possible to have one of the devices' IRQs share with the other, however this is all determined by the BIOS's DMI these days (in a modern PCI BIOS at least). I'm only talking about PCI here, ISA is already a forgotten issue since I'm talking about the latest and "greatest" motherboard.
Aren't they trying to keep some ancient inferior, simple interface up to date and competitive just because its "cheaper"? AFAIK, it should cost no more to make a SCSI device/drive with the *same* MTBF rating as an IDE device. IDE works, only when you are keeping things *simple*, but things aren't so simple these days. The more expensive, branded, prebuilt *gasp* systems these days already come with a decent sized HD, with DVD, and usually a burner, and sometimes a Zip or LS-120. This means 2 IDE channels may already taken up. IDE seems cost effective, but it doesn't look like it to me when it comes to long term. Its more trouble than its worth when you are going to add cards into your slots. Doesn't this remind you of the saying "beating a dead horse" to you?
It all comes down to this: we all know that we are in a serious IRQ resource problem already, and adding to that we get "newer and better" IDE "standards" which contributes to this problem even more. What I think should be done is to either ditch IDE (it worked great as a cheap solution but is no longer really viable), or take care of the IRQ problem. However, there is one thing that seem to be preventing this: the industry thinks they need to maintain backwards compatibility. I feel that there will eventually come a day where someone out there in some company will crack and actually officially acknowledge of this problem and is actually willing to deal with it.
My strongly suggested action is to actually make SCSI cheaper (man, they make tons of money selling those things, when costs of manf are no more than IDE), thus allowing IDE to be rid of, and in turn allow us to connect at least 15 devices (Wide SCSI) and using only 1 controller, 1 IRQ, 1 port, and lower CPU usage tremendously.
I still have to admit that IDE is ideal for people, and some of the geeks out there who are poor and can't afford good stuff like SCSI. But the minute you can afford and want to do serious (workstation/server) stuff, there is no doubt about it: SCSI is the way to go.
TheMAN
SCSI allows many things that IDE doesn't do. It has a rather mature RAID controller market, not to say IDE RAID doesn't exist in hardware and software. Also, I don't remember the last hot swappable IDE drive that worked. Since I deal with many arrays of disks chained off a smaller number of computers, it would even be *possible* to have the number of discs we have. You could only get four drives in an array with IDE. That's nowhere near acceptable. Plus since you can't do command queueing and simultaneous transactions on IDE, the performance is crap for it. Plus, since the cables have length limitations, especially for ATA/66, it's no good for external bays. IDE is nice and all, but I would never rely on it.
As for WD, just had *two* drives fail within five hours of each other from them. One a year newer than the other. Don't miss them a bit, perfectly happy with Fujitsu and IBM right now.
IDE? No thanks. I'll stick with my 10,000rpm Seagate SCSI disks (over 20mB/sec!!), nice 7200rpm IBM disks (12-16mB/sec!), and my Symbios dual channel UW SCSI.
Move over 20mB/sec from one drive with IDE? Perhaps if you get real lucky. Move over 80mB/sec from 4 of them with only one card and hardly any CPU usage? Never. Hook up a 12 bay SCA hotswap enclosure to an IDE controller? Not a chance.
He used to do long rants (if it is him) about "Sun and HP should just acknowledge that Linux/Intel has won, and absolutely smokes them everywhere, everyhow".
"Linux, Intel and EIDE are what power the Enterprise Storage/RAID Market"???
Open Source. Closed Minds. We are Slashdot.
Does anyone know if the current 1394 drives that are essentially IDE with a bridge to 1394 have the same CPU time consumming properties of IDE?
I really don't see what the big fuss is about SCSI starting to fade out when there is a superior standard in place. What really needs to happen is for true IEEE 1394 based drives to come out. The IDE bridge chips are starting to become the speed bottleneck, giving the bad impression that the 1394 drives can't perform.
after watching the transmeta webcast, I'm suddenly more conscious about power usage - remember that ATA hard drives control the laptop market (you don't see any 2.5 inch scsi hard drives, do you?) and ATA hard drives generally use less power. (I remember people telling me a computer with many scsi hard drives generally need to start the hard drives one at a time just to not overload the power supply) Oh, and scsi hard drives, especially, the 10000 rpm ones, emit a lot of noise and heat (requires some kind of cooler - more noice) whereas ATA hard drives get warm at most (leiz feels his IBM 7200 rpm hard drive, did not burn his finger)
.....
_______________________________________________
There is no statute of limitation on stupidity.
If SCSI sucks why did my DEC Alpha server come with only SCSI control and not IDE. You know not what you speak of, IDE is for single luser systems. SCSI is for serious systems
in the last 2 years, couple of my firends had a total of 3 Western Digital drivers die on them (they were the 4.3gig and 6.4 models) western digital replaced them for free however. at work, we do have a western digital scsi hard drive... had no problems so far in 1.5 years.\, which is a very short time. on a side not... i have a megaraid 428 and three seagate barracudas (32171 ultra wide models @ 7200 rpm) in a raid 0 config... i know hdparm is not accurate for testing hd performance, but hdparm reports only 11.6 megs/sec. it SEEMS slow as well... what is the verdict on seagate drivers? should i sell my raid config and pick up a ibm 10000 rpm lvd 2 drive?
Just like Operating Systems, word processors, and woodworking tools, you use the best tool for the job at hand. If you need fast, reliable disk subsystems designed for multitasking systems, go SCSI. If you're building an MP3 Jukebox for your home entertainment system, save $500 and buy a 30GB EIDE disk.
Personally, I have a dual Pentium Pro machine with all SCSI (cd, 2 Quantum Fireball 4.5GB UltraWide disks, Exabyte 8500 on a Mylex BT958). I have a single-processor machine with a single EIDE 16gb IBM disk & cdrom. Even booting the SMP box with only one processor, the UWSCSI disks running on a wide channel blow away the EIDE disk for raw throughput. But I do a lot of disk-intensive work on that machine, whereas the IDE machine is turning into an MP3 cache.
In short, when raw performance counts, go SCSI. If you don't care, and just want cheap space, go IDE.
I used to think printing on on Unix sucked. Then I figured it out. Printing on Unix *does* suck. Like a Kirby.
I've read a ton of stuff on the debate about SCSI vs IDE...and I've seen some people comment on how "SCSI seems to last longer" and I've seen other people comment on how "SCSI can handle multiple requests better"...but I must confess, it took an electrical enginneer to explain the reason that SCSI blows IDE away in servers, and always will: Let's start with the fact that most SCSI & IDE drives are identical in the hardware, it's the logic board that's usually different. Both the SCSI drive and the IDE drive have the same MTBF. Which drive is going to fail first in a server? The IDE will, every time -- BECAUSE IT WORKS HARDER, and RUNS MORE. SCSI's ability to get multiple packets of data means the moving parts of the drive don't have to work as hard as the IDE drive, which is sending the head flying over the platter for every little bit. Result? Two servers, same workload, one with an IDE drive, one with a SCSI, both drives have the same MTBF...but the IDE drive is chugging away to exhaustion while the SCSI drive caches some of the data it needs and is not working nearly as hard. This is why the speed debate is useless as applied to servers. In a desktop? Sure, IDE has its advantages, and big speed is always nice. But in a server in a business environment with a heavy workload, time is the value, and downtime costs -- and that IDE drive is GOING to fail because it's working 10 times as hard as the SCSI drive is to get the same data. Now, if someone can just explain why it costs so much more. I am inclined to agree with the previous poster who said that the hard drive companies just milk the "business market" but I have no real facts to base that on. Steve
But we won't tell if you share some.
Because of how universal it is. I got a SCSI CD-R because it was supported under linux. I got a SCSI cheapo scanner (Umax Astra 1220S) for the same reason. And once I got that hooked up, I could hook up this old exabyte drive that I had lying around and I could do real backups. SCSI not only has the peformance, but it's very useful to have around. (I just need a cheap 2 channel card. I hate having to have 2 scsi controllers in my box. :-)
oh my god.. my daughter will grow up to become Hemos..
*shudder* I needed that visual image, thanks ;)
Hope that isn't a preview of what is to come ;)
---The proceeding comments were not paid for by the following advertisers.
There seems to be so much debate over IDE vs. SCSI. I use both SCSI and IDE (IDE for hard drives, SCSI for CD-ROMS and scanners). I think its time for a better interface, like FireWire.
IDE's problems are well known. It takes up more CPU than it should, it is an ancient technology, and you can only hook up 2 per bus (each bus takes an IRQ). You can't hook up external drives via IDE. You need to jumper them for slave or master. Ugghhh...
SCSI is better in some respects, but has its shortcomings. It is expensive. There is a maximum cable length and those cables are dang expensive! You need to set SCSI ID's which is a total pain and should be a thing of the past. Termination is one of the biggest hassles.
FireWire doesn't have any of these limitations or hassles. It seems to have the bandwidth it needs. If only there was an internal interface...
USB is not in the same class as SCSI. Heck, in terms of i/o throughput, USB is not in the same class as IDE.
Don't get me wrong, USB is great for a lot of things. I wish every digital camera had a USB port, etc. For moving relatively small chunks of data (50 megs or less), USB is a great way to do things.
But sheesh, that's 12 mega *BITS*, and it's a *SERIAL* interface. Divide by eight, then account for bus latency. Then compare it to the ATA-1 speed limitations. Yeah, a hard drive on USB would be soooo cool.
Firewire is interesting, arguably in the same class as some forms of scsi, especially in regards to price.
This is just like television, only you can see much further.
Here's a tip sent around our company concering tweaking IDE perf. Thanks to Andrew Tridgell for the info.
/dev/hda /dev/hdc
/dev/hda /dev/hdc /dev/hdc > /dev/null & /dev/hda /dev/hda /dev/hdc /dev/hda
This tip is useful for just about any Linux box, and is probably the
simplest way to significantly speed up your IDE based Linux box
without changing the hardware.
If you are impatient then just add the following near the top of your
/etc/rc.d/rc.sysinit (or equivalent startup script):
/sbin/hdparm -u 1 -d 1
/sbin/hdparm -u 1 -d 1
(and so on for any IDE devices in your system)
Now for a more complete explanation.
By default Linux uses extremely conservative settings for IDE. In
particular the default settings do two things that make IDE perform
really badly:
1) DMA is not used. That means all data coming to/from the hard disk
or cdrom is processed a byte at a time by the CPU. That is not very
efficient. With a fast processor that isn't doing anything else at
the time this can appear fast in simple minded benchmarks but it is
a big drain on CPU resources when you are actively using the
machine.
2) hardware interrupts are masked during IDE transfers. That means
that while a lump of data is being transferred to/from a IDE device
no other interrupts are processed. This includes interrupts from
other IDE devices, from network devices, from serial ports and from
mice. Your whole machine is effectively clagged up doing nothing
but waiting for a horrendously slow device to say "I'm done". Not
good.
If you want to see just how slow this is on your system then do the
following:
1) put a CDROM in the drive.
2) run the following commands:
hdparm -d 0 -u 0
hdparm -d 0 -u 0
cat
hdparm -t
hdparm -d 1 -u 1
hdparm -d 1 -u 1
hdparm -t
that shows you the hard disk speed while accessing the CDROM with the
default settings and with the improved settings. On my system the hard
disk speed goes from 3.8 MB/sec to 12.9 MB/sec. I've seen much bigger
changes on some other systems.
Even more importantly than the speedups is the fact that you will stop
dropping your PPP connection while doing cdrom transfers, and you will
be able to use your system while burning a cdrom without creating a
coaster.
You may wonder why the default settings are so poor. The reason is
that there is some rare hardware out there that corrupts data during
IDE transfers when you either use DMA or receive an interrupt during a
transfer. If that happens then the kernel should detect the failure
(in nearly every case) and fall back to the default
settings. Unfortunately after the auto-fallback you are still left
with corrupt data in your cache. Luckily systems that don't handle DMA
and unmasked interrupts are really quite rare these days so it is a
pretty safe bet to turn the options I suggested above, especially if
your system isn't from the stone age.
For more info and piles of options for fine tuning your IDE system try
"man hdparm".
I have a wd7000 scsi controller that kicks much ass, except for one thing: They sold the rights to adaptec, who immediately decalred it obsolete before any nt4 drivers were released. Which means, I can use its builtin real mode drivers for sucky performance on a win9x machine, or the crappy win9x drivers, and that's it! There's a linux project that hasn't been going very fast because ppl don't seem to have these any more. And I've had more WD drives die on me than any others. except for maxtor of course.
I just got some new drives at work, and want to share the benchmark to enlighten the world about the ever important and exciting SCSI vs IDE holy war. Well.... maybe I just want to brag, so shoot me.
/sec %CPU
Here's the bonnie results for the system. The drives are two IBM ultrastar 18LZX drives, 18GB, 10k RPM, about $580. The controller is the onboard adaptec U2W on a supermicro P6DBU. They are setup with a raid0 partition using the new raid 0.9 code, with kernel 2.2.13ac2.
-------Sequential Output-------- ---Sequential Input-- --Random--
-Per Char- --Block--- -Rewrite-- -Per Char- --Block--- --Seeks---
__MB K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU K/sec %CPU
1024 10695 99.6 60193 53.8 21373 42.9 11092 95.4 50926 29.6 508.3 3.8
60MB/sec output and 50MB/sec input. Touch that with your punny IDE drives. And remember, use a bonnie test size of at least twice your memory, eg. -s 1024 for a 512MB system, so you don't measure the linux disk cache instead of the drives.
7200rpm + U/ATA66 can sustain some wickedly fast speeds. For this reason I chose this on an Abit BE6 motherboard and cheap 7200rpm IDE drives for my cheap budget server at my cash strapped school.
I was astounded when I ran an hdparm -t (without cache disk speed test) and it reported 21MB/sec. This went well beyond my expectations from a little cheap IDE drive.
In situations where you only have one disk per controller (the Abit BE6 has two U/ATA 66 controllers), 7200rpm IDE can actually outperform SCSI based systems. (According to an article on Thresh's Firing Squad)
HOWEVER, SCSI still beats the heck out of IDE in reliability, speed and scalability in large and important jobs (enterprise solutions). The redundancy and failover protection of SCSI + raid controllers is not as reliable with IDE (it's possible with stupid human tricks). Don't even talk to me about software RAID. Software RAID is too CPU intensive. SCSI + RAID controllers can do all the failover, drive rebuilding and cool stuff without the CPU knowing anything about it.
Also, the extra bandwidth of SCSI shines when many hard disks are added to the fray. IDE has nowhere near the level of scalability of SCSI.
So basically, I highly suggest U/ATA 66 IDE for desktops, workstations and low budget servers. But for large and important jobs use SCSI.
Enable write caching and you're sure to have
filesystem corruption within a short time.
As a interface for external devices? Definetly. who wants to plug a external ZIP or CDR drive to parallel port?
For medium-sized servers SCSI is still the best choice. IIRC, ATAPI doesn't implement hot-swapping, and I haven't seen that many IDE controllers with RAID support.
-- Reality checks don't bounce.
I did a few hdparm drive speed tests a month ago and here's what I can remember from those tests.
/dev/hde
Abit BE6, 7200rpm IBM Deskstar 20.5GB
hdparm -t
pio - 2.60MB/sec
U/DMA33 - 5.60MB/sec
U/DMA66 - 21.20MB/sec
Sure SCSI seems to perform as well as IDE when meassuered on a single user, single tasking, single disk system.
Now try the same on a multi disk multi tasking/threading environment (say a large shell machine, or a busy webserver). SCSI will then totally blow IDE away.
*IF* you want to run a benchmark, do meassure the correct values *please*
Is it because SCSI drives reserve a large portion of their space for alternate tracks (in case there's damage to a track)?
In any case, given the number of platters in large HD's, it is easy for the HD makers to remove a couple - with SCSI, you're already expecting to pay more for less space, so they might be milking you just that extra bit.
Anyway, if you're a desktop user, there's usually no reason to go SCSI. If you have a server which is used for any non-trivial task (e.g. a bit of print spooling or lightly loaded file server), get SCSI.
SCSI is *SO* much better with IRQs. Look at it this way.
.5 of an IRQ per device.
/proc/interrupts
IDE: 2 IRQ's, 4 devices. Thats
SCSI: 1 IRQ, 30 devices. 0.03 IRQ/Device.
From
15: 263999 XT-PIC aic7xxx, sym53c8xx
PLEASE DONT LET THEM KILL SCSI!
-Yarn - Rio Karma: Excellent
Why is it that people burning CDs using IDE CD-Rs are always shutting down programs to burn CDs whereas I can play Quake3 or compile a kernel while I'm burning?
On a more serious note, I would really like to see some sort of benchmarks between IDE and SCSI drives of similar speed (ie. two 7200RPM drives with close seek times).
Why moderate it down?
For the same reason that my first comment got moderated down - it's OFF TOPIC. Perhaps you're having difficulty grasping the concept, but comments are actually supposed to be about the article they're attached to.
Not to mention that their target market is largely portable devices. Does a laptop need support for seven hard drives? What's that you say? NO? Astounding! I thought everyone wanted to pay MORE for a laptop!
External ports! Everyone in the world is obviously lugging a huge scanner around with their 5-lb. laptop. So let's build all our laptops to suit a market of 800 people and try to sell them to a market of 50,000 people!!! It makes perfect sense!!!#$%@#!$#$%!#$%;LKAHSDF;LASKDG;L
This site is "News for Nerds", not "Endless navel-gazing about how to run the site". If you want that sort of discussion, go hassle on of the infinite number of sites devoted to pointless flamewars about how they should be run.
I was reading the comments, getting angrier and angrier with the price difference between IDE & SCSI, when I thought this: 'I wonder if it would be possible to rip of the IDE controller board from a hard disk and replace it with a SCSI one'
Any thoughts?
-Yarn - Rio Karma: Excellent
being the purist that I am, alpha still holds more appeal for me.
running a GUI on an x86? bah!
running floating point calculations on an x86? bah!
I like having one processor for everything. a nice clean instruction set. no stupid 32 bit overflow problems.
alpha makes sense.
(moral of the story: price matters.)
While you are right that narrow/wide has 8/16 target ID's, that does not neccessarily mean you can plug 7/15 devices onto a SCSI bus.
Eg, Ultra will not really tolerate more than 4 devices on a bus. It's possible to have 5 or 6 Ultra devices on the same bus, but you are likely to have problems.
I use Friend/Foe + mod-point modifiers as a karma/reputation system.
scsi wuzzy was a bear
...I mean, doh! I mean, woohoo! I've been nothing but happy with my WD Enterprise, so it irks me that they won't be making any more, but maybe now I can pick up some cheap drives at firesale prices. w00t!
Never approach a vast undertaking with a half-vast plan.
Western Digital's caviar models from 1.0GB up to 4.0GB are notorious for failing. At my company we use Gateway computers, which have WD, quantum, and maxtor drives. The WD's have a high rate of failure, such as developing bad sectors, and of course the Clunk of Death, where one day you turn the machine on and the drive just sits there going "clank clank clank clank clank..."
Now, the new Expert line is a different story. They're licensed from IBM technology and should be just as reliable.
-CausticPuppy "Of all the people I know, you're certainly one of them." -Somebody I don't know
Last time i checked ide had difficults handling more than on read/write request at a time. Not very sufficiant for a multitasking os trying desperately to write multiple sectors at once?
Uh, cached accesses would run at 33MB/second or 66MB/second depending on the IDE bus speed.
Reading and writing a 100MB file (i.e. far, far too big for the cache) on a crappy inefficient FAT32 file system under crappy inefficient Windows, my 36GB IDE drive sustains 18-22MB/second; I don't have a disk test program for Linux, but it would probably be even better.
SCSI is certainly better for servers where you need to access multiple files simultaneously, but for normal desktop use IDE gives you similar performance for half the price.
They make excellant scsi cards and have well documented hardware specs. Not to mention their linux drivers. Plus they are half the price of adaptec.
why?
As I said above, my UDMA33 5400rpm 36GB Maxtor drive sustains 18-22MB/second even under Windows. Why run a hot, expensive 10,000rpm SCSI drive when you can get the same performance out of a $250 IDE drive?
Yes, SCSI is better if you need lots of drives (I have a SCSI CD-writer and DVD-ROM on my PC to free up IDE slots for hard drives, so I'm not an IDE bigot by any means), or you're running a server, or you need lots of drives. But for most ordinary users, IDE is at least as good.
And I would add that finding the right adaptors, terminators, etc, etc to connect those SCSI devices to my SCSI controller took me about a week (and the adaptor and cable to connect the DVD-ROM cost 1/3 as much as the drive!!!!!); most IDE drives just plug in and work. Again, for ordinary everyday users this is a big benefit of IDE over SCSI.
You have never had a Fujitsu crash? Try driving them around in your car. I have had three go during transportation and 2 more just go for no reason. .02 Euro's....
Everyone should buy a good brand , pay the extra 5% for it and sleep easy.
Just my
ROTLFMAO... a SCSI bigot complaining that IDE cabling is complex! Let's see now, exactly how many different kinds of SCSI connector are there? A dozen? Two dozen? How many times have your drive refused to work because of some bizarre termination problems?
Compared to SCSI, getting an IDE drive up and running is a breeze... But yes, you're right that they really, really need to do something about the limited number of devices it supports.
A modern two-drive IDE RAID system will give you 40-50MB/second, and you'll get two 40GB drives for the price of one of your 18GB SCSI drives...
So for the same price you get 36GB running at 50-60MB/second, and I get 160GB at 40-50MB/second... four times the space and only 20% slower. As a video editor, I'll take that extra 120GB, thanks.
In my experience, an IDE CD burner enjoys making coasters if anything (even a screensaver) decides
to share the computer with it. (My experience is with a Celeron366/Win95/Mitsumi IDE CD burner.)
I have never used a SCSI system, so I don't know
what difference it would have.
I personally have had five Western Digital IDE drives go bad, especially that one series (1.2 GB or 1.6 GB?) that Western Digital couldn't make reliable to save their life.
To their credit, they replaced all of the drives I've had go bad (one even went bad in the first six months). But I don't even use the replacements, one of them is still ni shrink wrap even, because I don't trust my data on a Western Digital drive.
The company I work for just bought the "brand new, never been used" large research facility which WD built, but hasn't even completed construction on yet. They took a big loss on that, so you know something serious is up. Lots of people at WD here are expecting to lose their jobs.
After reading a lot of WD sucks posts here I thought I'd chime in with some of my personal experience with them...
First off, I'd love to be using SCSI. But its just too expensive for my budget (wich is not exactly small, I just replaced a 2yr old P-II with an athlon sys)
On to disks... I've used lots of WD disks. The first machine I owned, a Quantex P-100 I bought about 6yr ago came with a 1.2G WD caviar (mode 4). I later added a 2G Seagate Medalist (mode 4) to that box, but the WD was always a bit faster, albeit louder. FYI that box was running dual boot Win95/Slackware (1.2.1 kernel). Today that box is used by my little brother and the WD is still running strong! And that drive was up 24/7 for the first 3yr of its life!
Second box I built was a P2-266 that I powered with a Seagate disk. I later added a 8.4G WD caviar ata-33 disk. 2yr of 24/7 uptime and that WD is still doing fine as well.(its now a dedicated RH 6.1 box)
Along the way I also built boxes for my father and a couple of friends. The only disk problems I ever had were upgrading my fathers machine. The maxtor disk he had would not live on the same controller as a Seagate. And that Maxtor later died, only disk I've ever seen go bad.
Newest box is an Athlon I just put together using a WD Expert 27G disk. Runs great and the disk is very quiet.
If I could choose any disk, my first choice would be an IBM deskstar, but they are hard to find and a bit pricey. I used to really like Seagates but they seem to be falling behind lately. I had trouble with maxtors so I stay away from those. That leaves WD. 6 years and few disks later I have no complaints.
Jeremy-JeremyH
>EIDE will eventually hit limits as even desktop computers become more demanding.
But will this make a difference? Most people/companies/tech guys are complaining about how other componets of a computer have to be faster. Bus speed, CPU speed, video card, network/Internet connection are all mentioned in various non-database related articles but I have yet to see disk speed being a factor.
For me the most important part of a HD (besides its capacity) is its noise. HD are fast enough for me and most of my applications but I have a friend who just got a SCSI HD and that thing sounds like its crushing rocks.
The surprise isn't how often we make bad choices; the surprise is how seldom they defeat us.
That's because there's nowhere where it would be on-topic. But I guess we should only dicscuss approved topics, huh?
God, don't try thinking for yourself anytime soon.
I never do this but I have to today. Thank you for putting up with me. Argh! I am so friggen sick of idiots! Quad Xeons faster then 30proc Sparc 6500?!?!? IDE Raid outperforms FIBRE and SCSI?!?!? Where are these people comign from? What friggen benchmarks have they run? Who told them this crap? Ahh, ignorance must be bliss. I want the drungs their on...
Folks,
I think many of you are missing the point.
The big advantage of IDE is simple: low cost. Remember, in the old days you had to buy a separate hard disk controller, and that hogged valuable expansion slot space (not to mention the time wasted in doing a low-level format of a hard drive.)
Since IDE drives don't need a separate controller card (and don't need low-level formats), all you need to do in 1999 is connect the drive to the motherboard (heck, even the system BIOS will automatically set up the drive type), and you can right there install the operating system of your choice.
Also, in the past people have rightly criticized about IDE drive's low performance compared to SCSI drives. However, with Intel shipping the 82371 series of I/O controller chips, that allows software drivers to be written that dramatically reduce the CPU utilization to access an IDE drive. Also, the development of Programmed I/O Mode 4 in the early 1990's, ATA-33 in 1996 and ATA-66 in 1999 has dramatically increased throughput on IDE hard drives to the point that for most desktop operating systems there is almost nothing to be gained by going to SCSI hard drives.
The only place where SCSI hard drives still are useful are in environments where hard disk access is very heavy, such as in servers. This is where the RAID 5 capability of modern SCSI host adapters and the throughput of SCSI Ultra-Wide and Ultra2-Wide becomes useful.
It's small wonder why Western Digital is no longer interested in SCSI hard drives. That's because IDE hard drive technology has advanced to the point that SCSI hard drives are only useful for server environments.
Raymond in Mountain View, CA
Did you even pay attention at the all hallowed presentation yesterday? There don't sound like there are going to be any desktop Crusoe motherboards, at least not in the immediate future. It's technology that's tuned for portable use.
And if you're going to try to convince them to adopt 'mandatory SCSI, no IDE' on their hardware, why not advocate solid gold circuit traces on the PC boards? Wintel doesn't do that either, and it's about as stupid as any other form of arbitrary 'spurning' activity on the part of a competetive business.
SCSI Fanatic: I can burn 2 CDs while I play quake 3, do 24 bit colour scanning, leech mp3s and watch the latest pamela anderson vid.
IDE Fanatic: IDE is cheaper.
SCSI Fanatic: Oh yeah????? Well my drive rotates at 20,000 gigaschmirkels per second and I can chain ***37*** DRIVES TOGETHER!
IDE Fanatic: IDE is cheaper.
SCSI Fanatic: So, my mega-ultra-fat-wide-giga-fast-scsi-4 drive can do simultaneous reads and writes and can reorder requests fast enough to pilot the space shuttle - LETS SEE YOUR IDE DRIVE DO THAT!!!! AHHAAHAAHAAHHAHAHHHHAHAHAHAHHAHAHAHAHAHAHA AHAHAHAHAHAHAHHAHA AHAHAHAHAHAH AHA!
IDE Fanatic: IDE is cheaper.
Do you see what I am getting at? Nobody is saying IDE is technically better, yet all these wonderful SCSI fans are screaming until they turn blue in the face and the veins in their forehead start bulging out. Frankly it's disgusting.
Stop it.
I mean it.
I can still see your veins.
I wasn't joking.
I can tell you why they stopped making them. It has nothing to do with which is better (IDE/SCSI/FC-AL) or anything. What is has to do with is that WD makes HORRIBLE SCSI drives!! they are the WORST quality hard drives I have ever had the mis-fortune of running across. I'm still using 5-6 year old seagate drives and they still run perfect, but i'll be damned if I have a WD SCSI drive last longer that 6 months! My company bought six of them for our main file server and 3 months later 4/6 were bad!
Take it out and wave it around.
The Deluxe Dicksize wars. Brought to you by Slashdot.
...is that even though they keep updating the technology, they don't keep changing the damned connectors.
How many different types of SCSI cable are there, between original SCSI, Wide SCSI, Ultra Wide Scsi, etc.?
I can still use the same 1GB drive that I had on my 486/33, on my much newer PII/450. And I can use the same cable. This makes me happy. The fact that IDE is also cheaper than SCSI makes me happy.
Now, if the SCSI advocates are only talking about practicality for high-load file servers, and server candy like hot-swappability, they have a point. But if they want to treat workstation practicality as being equivalent to server practicality, that's an all too common fallacy.
Terrorists can attack freedom, but only Congress can destroy it.
The future is in serial buses, not in SCSI, both for the big server markets and the luser market. Firewire is already there, it might not be the standard though, given the lack of Intel chipset support. Why are serial buses superior? After all, the more wires you put, the more data you pull, don't you? Not quite: synchronising those 2^n lines is a bitch. Also, as the price go down, it will make low-tech parts (copper, plugs, pcbs) less expensive.
if i recall correctly, i think usb has a max transfer rate of 12Mb/sec (being mega bits) which is approx 1.5MB/sec (mega bytes).
now imagine a harddrive running at 1.5MB/sec. thats just a little bit better than an internal ide zip drive
this is why firewire was developed.
john
-- john
Well Well Well...
If you take a look at the guestbook to the Mr. T Site, you will find this entry
Name: Craig 462er5ggd2g3feww McPherson
Website: Lacey Chabert Meets Pokey the Penguin
Referred by: Just Surfed On In
From:
Time: 2000-01-19 19:57:05
Comments: Hello, I am currently looking for a method of transforming cute teenage girls into stone statues. I really need help with this. If you have any hints, tips,
or ideas with regards of petrifiying nice young ladies, please e-mail me. Peace out.
By following him to his website:
his email address *may be*:craig@laceyonline.com
I THINK I'VE FOUND YOU YOU F'SKER.
I just go a box with an Abit/BE6 board, it indeed has 4 IDE channels, but two of them are UDMA/66. They seem to be a bitch to use on Linux. How do you use them?
WD sold off the high-end disk drive controller division to Adaptec 4 years ago or so. (The group was then sold to TI...sheesh). I'd guess that most if not all of their expertise in SCSI went with the sale (personal note -they are GREAT engineers, I had fun working with them.)
Have you compiled your kernel today??
One word: Firewire.
--
This space unintentionally left unblank.
Moderation Totals:Flamebait=1, Insightful=1, Informative=1, Total=3.
It's listed as (Score: 0,flamebait)
I don't get it?
0
+1
+1
-1
=0?
I wanted to point out that with ACPI, Automatic Configuration and Power Interface, the IRQ issues all go away, I see it as an invalid argument. In my ACPI systems I have my SCSI card, AGP Card, VGA Card(yes, multiheaded), NIC, SCSI, Sound, and USB all running on a single IRQ, IRQ9. I don't know if 9 is significant as the 8bit-16bit bridge w/IRQ2 from the old 286 days or not. But the point is, IRQ problems are a thing of the past.
Just like any technology, each has its place. My grandfather who wants 20Gigs of space to make home movies on his iMac DV or PC workstation shouldn't have to spend $500 for a disk. He doesn't. In fact the space can be had for under $200.
On the other hand, a profession movie maker will want a nice set of striped 10krpm SCSI disks to really be able to read and write all that movie information, these are the workstations that get RAID arrays and are well justified.
I don't see the conflict myself. I would love a SCSI system, but my pocket books doesn't. Right now I'm more than satisfied with my 7.2krpm IDE drive running 11gigs of Reiserfs. It is actually the second fastest filesystem I've ever worked on next to Compaq Dec Unix's Advfs on 5 IBM 18Gig 7.2krpm disks in a RAID5 configuration.
That said nothing comes close to SGI's disk arrays. So much efficient striping allows for over a TB/s sustained transfer rate. But looking at the $$$ it is believable.
I use SCSI exclusively in all of my systems, I even have an ancient 286 with a Seagate ISA controller and Quantum fifty-meg drive.
As to the complaints about SCSI drive ID issues, newer drives and controllers (Adaptec 2940, etc) support SCAM (SCSI Configures Auto Magically), which automatically handles drive IDs for the user.
I prefer the Sun solution- SCA drives, one single connector on the drive for power and data, with the drive bay determining the SCSI ID. Clean and simple.
I do not deploy Linux. Ever.
[*] For those who don't know what this is. Suppose we have three files spread over the disk like this:
[--A---B--E--------------C--------------------D--- ]
On Windows and most single user OSes, a SINGLE program typically does things sequentially, so the order it reads the files A, B, C, and D in is less important. On a multi user or multi process or SMP system, simultaneous disk read requests can come from simultaneously running programs each doing totally different things. Let's suppose the SCSI drive is just finishing reading file E. Now, if the read requests to the SCSI bus come in for files C, D, A, and B, in that order, the SCSI logic will recognize that the drive head is nearer to files A and B and that it would take longer to do the reads in the requested order. So SCSI will reorder the requests and read A, B, C, and then D, thus saving lots of time by not swinging the head from one end of the platter to the other. This makes SCSI *much* faster and efficient. With IDE, when a read/write request comes in, the IDE controller sits and waits until the request is completed before it allows the next request to be processed. THIS IS LAME! And this is why IDE is much slower on real computers.
I purchased an AC31600H (1.6 GB ATA) about four years ago. After about six months, it started having "knocking" symptoms. WD's standard response is, disconnect it from the cable. If the symptom stops, it's your problem.
No other disk drive in my possession ever manifested that symptom. When WDC issued an advisory about those disks, I requested a replacement. The replacement suffered the same symptom. I got them to replace it once more. They sent an AC21600H. It ALSO had the symptom. I gave the disk away and haven't looked back.
In my experience, WDC's best disk was the ST-238 (30MB RLL).
I will _never_ buy a WD disk drive again.
My Maxtor, Conner (before Seagate bought it), Quantum and Seagate disks perform flawlessly.
I have a Conner 420 purchased at least six years ago. It still runs fine as the Linux "/" drive of my wife's computer.
No more WDC.
(original post was in the wrong place)
I've read a ton of stuff on the debate about SCSI vs IDE...and I've seen some people comment on how "SCSI
seems to last longer" and I've seen other people comment on how "SCSI can handle multiple requests
better"...but I must confess, it took an electrical enginneer to explain to me the reason that SCSI blows IDE away in
servers, and always will: Let's start with the fact that most SCSI & IDE drives are identical in the hardware, it's
the logic board that's usually different. Both the SCSI drive and the IDE drive have the same MTBF. Which
drive is going to fail first in a server? The IDE will, every time -- BECAUSE IT WORKS HARDER, and
RUNS MORE. SCSI's ability to get multiple packets of data means the moving parts of the drive don't have to
work as hard as the IDE drive, which is sending the head flying over the platter for every little bit. Result? Two
servers, same workload, one with an IDE drive, one with a SCSI, both drives have the same MTBF...but the
IDE drive is chugging away to exhaustion while the SCSI drive caches some of the data it needs and is not
working nearly as hard. This is why the speed debate is useless as applied to servers. In a desktop? Sure, IDE
has its advantages, and big speed is always nice. But in a server in a business environment with a heavy
workload, time is the value, and downtime costs -- and that IDE drive is GOING to fail because it's working
10 times as hard as the SCSI drive is to get the same data. Now, if someone can just explain why it costs so
much more. I am inclined to agree with the previous poster who said that the hard drive companies just milk
the "business market" but I have no real facts to base that on
Hey guys got a question...I did get a new 27.3 gig WD DMA ATA 66 drive for Xmas. Ive got the drive holding Linux , NT and Win98. I didnt try to enable DMA on it yet in NT or Linux however I did in 98. 98 turned it back off after having trouble booting the operating system. My mobo is a DFI Intel Triton TX? chipset. I dont think the board even supports ATA33 I dont ..think but still it should support DMA as I have my CDROM and my 6.4 Hd on DMA both. Anyone have a clue why yhe 27 gig would not want to have DMA enabled. BTW the CMOS reports the 6.4 as UDMA mode 2 and it reports the 27 gig as UDMA Mode 4. If anyone knows anything Id like to hear from you. nicole@NOSPAM.magenet.com
The reason normal people (myself included) can afford decent machines and good bandwidth is BECAUSE OF Joe Bob; if Joe Bob wasn't willing to buy, costs would be astronomical, and computers would be limited to corporations and universities.
Yes, SCSI is wonderful; yes, IDE sucks in many ways; yes yes yes. But IDE is good enough for most people, and it's cheap, and it maintains backwards compatibility, and for 90% of everyone, that's what counts, and that's what drives the industry.
-brennan
Ok, here are some order of magnitude data access stats.
Lets say we have a 500MHz CPU that can process one instruction per cycle and it needs to get hold of a bit of data. How many cycles get wasted?
Register : 1 cycles = 2e^-9 seconds
20ns RAM: 10 cycles = 2e^-8 seconds
10ms SCSI HD: 5000000 cycles = 1e^-2 seconds
15ms IDE HD: 7500000 cycles = 1.5e^-2 seconds
Soooo...
Do you spend £500 on a 700MHz PIII, on 256Mb of RAM or on the fastest god damned disks you can find?
Deleted
This ought to give any SCSI system a run for its money, especially as Firewire supports 63 devices and even some current Macs have 3 Firewire busses.
Firewire drives are starting to come through from several manufacturers, ranging from expensive high end drives to more cheap and cheerful.
Mac OS 9 users report being able to unplug the cable on a firewire drive *in the middle of a file copy*, and then plug it back in and have the copy complete. (The manual tells you not to do this, but it does work).
Firewire products guide:
http://www.apple.com/firewire/f irewireproducts.html
Lord Pixel - The cat who walks through walls
Lord Pixel - The cat who walks through walls
A little bigger on the inside than out
This is actually not true. Highend IDE will often share the same HDA as the low end of SCSI. Example being the Barracuda IDE drive. There are no Cheeta or Atlas 10K IDE drives. Manufactures need more than just the interface to differentiate between their lines to justify the price diff.
So the IDE adage is basically true: IDE IS AS FAST AS SCSI FOR 1 DEVICE. However, CD-ROMS/CD-Rs/etc count too.
Just FYI, I have a 9G U2W IBM disk, a CD-R, a DVD, and a Jaz all hanging off of a Adaptec U2W controller, and the system handles heavy IO very smoothly. That's the only excuse I need to spend a little extra for SCSI.
- Robert
IDE is to SCSI as Winmodems are to real modems.
I run a system with mixed IDE and SCSI, as I imagine a lot of the people reading /. do. SCSI for the OS and applications, and a huge IDE for data, and less frequently accessed files, as well as my collection of legal mp3s. However am looking to add more SCSI storage to my system, but do not have room to ad another SCSI drive inside the case. Does anyone have suggestion or warnings about external SCSI enclosures.
IDE RAID devices kludge around the two-device-per-channel limit simply by having a whole slew of channels on one board, and the same applies to the boards that claim to support eight or more drives. My point is, why bother? By the time you get to that level of complexity, you're either at or beyond the same level required for a good SCSI host adapter, and you still don't have the advantages of a true bus.
IDE is fine for a couple of drives, especially if one is on a tight budget. But if it's flexibility in combination with low system overhead that you're looking for, SCSI is still (and will continue to be) a solid option.
Oh, and SCSI drives need not cost a fortune either. There's lots of good deals on the surplus market.
God, don't try thinking for yourself anytime soon.
I never thought I'd see those words from the mouth (figuratively speaking) of an AC...
I seem to be missing something. Where is Natalie Portman mentioned?
I think you need to check yo head, mutha fucka
Besides, if you value your CPU cycles, you won't waste your money on IDE. As for pee-cee users, IDE is about all they can understand (though I doubt most of them can spell it).
WDC drives are so bad, they make the news:
pcweek article from Feb 1997
cnet article from last year
I've got several MFM/RLL Hard disks at my place, mainly in XT class comps. They are good for the applications that I run on them. Sure, the speed is slow, but they work well. I like SCSI because it's easy to install on comps. Sometimes the Stats for HDDs when autodetected are wrong. there isn't any config probs with SCSI, because it uses firmware to controll drives. Win boxes have made SCSI server hardware, because IDE is much cheaper. SCSI will generally run at least %25 or more than IDE. SCSI is better, but I'm 16 and strapped for cash, so I'll take what I can get. Maxtor is supeiror anyways, but Seagate makes good drives. I'm running two medalists, and am waiting to upgrade. You can have many more SCSI hdds on a system, anyways. (I heard a guy talking about having capability to put 42 on his, the bios would take forever to use) This makes it much more pratical for applications such as net servers. SCSI will become more popular as the OSes get larger and larger, because IDE will take longer to load. IDE is what I can get for $100 for 17 GB (The advertised them at Staples, but they don't have it in stock), so I'll be buying one of those if they have it in before the rebate runs out.
Don't call my crazy, that's what they called me back in the home!
I will not repeat the many valid resons already sited as to why SCSI is better than EIDE, but I will mention some new points.
Credentials: 2.5 years as a test engineer at Seagate Technology in Scotts Valley, CA.
Many of the EIDE drives I worked on claiming to be EIDE 16MB/s capable were indeed capable of delivering data at that rate, but could only sustain the rate for ~ 20 to 24 bytes! After 20 bytes the drive would raise the IOCHRDY signal, indicating it could not deliver any more data. But a logic analyzer confirmed that those 20 bytes WERE delivered at 16MB/s! I'll bet many other EIDE drives on the market have the same limitations.
SCSI has all of those expensive terminators because at those data rates the SCSI interface cable has significant impedance and transmission line effects can no longer be ignored if data integrity is desired. EIDE is facing the same problemsat the current rates, and drive manufacturers have reccommending limits on cable length to 16 inches. UDMA (a.k.a. is simply a hack to double the data rate without doubling the clock rate. In UDMA the IOR and IOW strobes run at the same rate as EIDE, but words are strobed into/out of the disk interface on both the rising and falling edges instead of just the falling edge as in standard ATA. The EIDE drives are NOT going to get another performance boost without increasing the clock rate, and that will require terminators. Oh, my!
Hey folks, lets face it, it way past time to phase out both IDE and SCSI and replace them with a modern standard such as IEEE.1394 (FireWire, iLink). 1394 has many advantages over both systems:
Hot Pluggable!
No ID's, IRQ's or other stuff to mess with
No termination issues
Power available from the host (except in portable situations)
Way faster. 400mbit/sec now, 800 almost out, gigabit speeds in the near future
Many more devices per bus. Topology can be daisy-chain or branched with hubs.
The bus is useful for much more than just storage devices. Literally millions of digital video cameras are already deployed with 1394. It is also useful for digital audio and as a LAN. (to be fair, SCSI is often used for scanners)
Very long cable lengths, over copper or fiber.
Cables are thinner, have less wires, and should be more reliable.
Connectors are smaller and tiny miniature connectors are available, which is why it fits on video cameras, where space is at a serious premium. The diminuitive Sony Vaio has 1394 on board!
Summary: IDE and SCSI are crufty. Something much better is available. Let's use it, shall we?
Dr. Burris T. Ewell
I guess I've just not been paying attention. WD dropped SCSI several years ago. I didn't realise that they had come back. Now I guess they're dropping SCSI again. Oh well. They obviously weren't much of a contender in the SCSI market anyway.
Many people who would presently benefit from a SCSI drive do not have one. If more of them did, SCSI prices would come down and they would make financial sense for even more people. This will never happen unless there is a widely accepted, easy to understand, single benchmark number. This number would have to encompass all the different specs. that effect HD speed in a weighted average for some broad range of uses.
Whatever weighted average was chosen it would obviously not be true for all situations. It would be better than bandwidth and RPMs or nothing as most people use today. The full specs would still be there for those of us who care. The next step would be to get all the manufacturers to use this benchmark. This should be as easy as getting a peace settlement in the Middle East.
db
I've owned several HDDs in my life. 4 were/are WD, that I can remember:
.WAV file. That's the difference that SCSI makes to me, not the theoretical transfer rates of the busses.
1) A 320mb monster (at the time). After several years, developed a few bad sectors.
2) A 1.2gb drive. This one is currently playing MP3s in the back of my 3000GT (which, for those that know, isn't exactly a smooth ride) with absolutely no bad sectors and no complaints.
3) 2 8.4gb drives. Both happily running, one in a Linux server and one in a (mostly) gaming system.
4) 1 10.4gb drive, in a gaming system, no problems.
I'd say I've been very satisfied with my WD experience, but then I've only been dissatisfied with a HDD once (a Fujitsu POS). I've owned Quantum drives and Seagate drives, and had no complaints about either. I used to work as a tech for a (once) very large (one of the top 3) PC manufacturer. This company used WD, Maxtor, Connor, Seagate, and Fujitsu drives in the systems. The Maxtors were notoriously crappy, as were many of the Connors.
The HDD that I love the most, and have been the most impressed with, and would buy again the most readily, is DEFINATELY my IBM Deskstar 9.1gb UW drive. It's a DNES drive (for those that know), and I absolutely love it. You IDE-lovers out there just try to rip and encode (rip directly to MP3) a CD from an IDE CD-ROM to an IDE HDD in AudioCatalyst without moving the CPU-Usage bar a SINGLE NOTCH! I do it all the time from my SCSI CD-ROM to my IBM. I can do anything I want while I'm doing it. On the IDE system I used once, I couldn't rip straight to MP3 (the meter stayed pegged and the "Possible Speed-Related Problems" counter was somewhere in the hundreds when I stopped it halfway through the first song) and I couldn't touch the system while it was ripping to a
Karma: Poor (Mostly affected by lame karma-joke sigs)
I guess I should address the Western Digital part first - they bolted a SCSI interface on a low performance drive mechanism and it (performance) sucked. I know this first hand - I ran into this while working on a fairly high (disk) bandwidth audio workstation project. The WD Enterprise drives couldn't sustain a transfer rate anywhere near the competing drives at the time (Quantum Atlas, etc.) My guess is that "I aint the only one" who figured this out, and word got around. People stopped buying the drives, WD wasn't willing to put the R&D into making a better product, and now they are exiting the market. No big loss from my perspective.
Now, as to SCSI in general - there will always be advantages. The IDE RAID cards mentioned are not "real" RAID cards - they just do striping and mirroring "somewhat" in hardware. A real RAID controller has onboard CPU and RAM and can do RAID level 5. Unless you can do level 5, there is almost zero benefit to RAID, IMO. On that note, you generally need SCSI for RAID 5 because usually you will be arraying a large number (eg 8 or more) drives to achieve high storage density. There are other advantages to SCSI as well - example: I do a lot of work in audio applications that involve manipulating large amounts of uncompressed data. SCSI works wonders in these applications - in addition to sever applications as other I'm sure will mention.
And as to the nut cases who claim they are getting 33MB/sec or more from a 7200 RPM U66 IDE drive - pass over whatever the hell it is you are smoking. Maybe you can get that kind of transfer rate by repeatedly reading a very small file that is stored in the drives cache or something. But in "real" applications like those I have to deal with, that involve moving large files, I find that the transfer capacity of a 7200 RPM drive is more in the range of 10 to 20 MB/sec, typically closer to the low end of that scale - SCSI or IDE doesn't matter here, that just what the drive is capable of.
This discussion is so old it reminds me of the old Mac vs. PC days.
I don't care what new ATA/DMA/Mode spec they come out with. IDE still unessarilly burdens the CPU and its performance is inferior SCSI. Only a complete overhaul of IDE could ever hope to solve this problem, and if they did overhaul IDE, it would no longer be IDE it SCSI in disguise!
See Apple McIntosh systems for a prime example. They have used SCSI hard drives since the early days. It's not that they think better. They just don't think "MONEY" all the time. Companies continue to push inferior IDE drives for financial reasons alone. Maybe that's why Apple has had financial troubles for sooooo long. :(
It's not that they think better. They just don't think "MONEY" all the time. Companies continue to push inferior IDE drives for financial reasons alone.
Maybe that's why Apple has had financial troubles for sooooo long. :(
I've never bought any WD drives because of those reasons. I've had a Maxtor UDMA/33 drive fail after one year, but they replaced it free. (SCSI is another matter--the Maxtor SCSI drive (no longer manufactured) in my NeXT is still going strong after ten years!). Seagate SCSI has also been rock solid.
"It's better to keep your mouth shut and be thought a fool than to open it and remove all doubt."
I mean geez. Talk about ease of use! Sign me up!
DO NOT DISTURB THE SE
Well, to me it comes down to this: SCSI is rediculously expensive. Though I myself have an AHA-2940UW controller, I bought it used for $100, not the usuall $250+ I have seen everywhere. SCSI cards and HD's have overblown pricing. There is nothing I can think of that justifies the high cost of SCSI Hard disks. The controllers don't cost THAT much to make, and so shouldn't cost as much to the end user. HD's cost even less to make. A SCSI HD should cost the same as an IDE drive of the same size. If they would do this, then more people would buy SCSI. I will use SCSI in any way I can because of it's performance (CD burning, true multitasking, overall faster), but I think this thing with WD will continue as a trend with other companies until either SCSI gets hurt bad, or the price drops to a more reasonable level.
I see and replace a lot of drives in my line of work and I have a preference for (in order): IBM, Fujitsu, Seagate. The others are much worse in terms of real MTF, especially Maxtor.
_damnit_
_damnit_
It's my job to freeze you. -- Logan's Run
Please moderate the parent post up for being "correct" and I/O secular.
But it was slower than my ATAPI ZIP-250 (1.10 MB/s vs. 1.67 MB/s)!!
But SCSI is amazingly reliable and fast. I've only ever had two scsi disks fail--one because of a power brownout, and one Conner disk (which I got a free Seagate replacement for, despite its being mfd. way back in '94). And although according to hdparm my new UDMA/33 disk is twice as fast as my 5-year old SCSI, the SCSI system feels faster (even though it's a slower processor!).
"It's better to keep your mouth shut and be thought a fool than to open it and remove all doubt."
The WDs I've had in the last 2 years have all been crap anyway. I get them replaced and they fail again within 6 months. Pathetic. For new PCs we ask for Maxtor or Seagate (and we've had a couple of bad Seagates recently). I'll drop Seagate and start trying Quantum.
I started CD burning on my mac with Toast. It FORCED you to do nothing else. (gotta love that cooperative multitasking ). Literarly. Win & Linux users won't know what I'm talking about, but a Mac app can completely take over. anyway, I got used to the same thing too.
When I got my IDE burner at home, I don't do anything while it's burning. sometimes I will read email, but that's only, like, once or twice. Coincidentally, I've never burnt a coaster with it.
DO NOT DISTURB THE SE
Yes, other things matter but HD performance, or should I say storage performance (hard disk plus controller) is one of the biggest issues that effect the "speed" of a computer right there with processor speed, amount of memory (RAM, L1 and L2 cache), and memory speed. All programs load from perm. storage. All files are stored there. All memory paging (or swaping) occurs there. As operating systems grow and commonly used data files grow (full length dvd quality movies) storage needs to be faster.
Drive loudness can vary with the design. EIDE has caught up but for awhile SCSI drives spun much faster. Also, the case and position of the drive can effect the noise. Open a box and you will hear how load EIDEs can be.
Storage performance becomes really important with database, application, and web servers. 100 Mb lan can push some systems storage such that storage performance becomes the bottleneck. Multi-processor, high Mhz machines push the storage.
This stuff really effects totaly computer performance for all systems from baby pcs to super data centers.
-- soldack
one of the direct impacts of this announcement is being felt here in Rochester, MN where WD Enterprise Storage Systems had just built a huge new R&D lab, and were preparing to plunder the staff of IBM's DASD group here in town... now they're going from moving into their new building and staffing up something like 600 jobs... to selling the building to Mayo and laying off the 400 some odd people that work at their current location. Bet those folks that left IBM are thinking about just how well they'd be welcomed back.....
I have used both SCSI and IDE based systems and I have to say SCSI fares much better in the performance area like anyone here could tell you. The big deal with IDE on desktop systems is back when a 1.2gb drive was enormous the drives on your PC were IDE. For hard drive manufacturers to keep selling their products they need to make damn sure their products work with legacy systems. So many years of IDE being dominant has left IDE the dominant standard today, SCSI is still too expensive for most people not to mention the need for a SCSI adapter. Motherboards with SCSI adapters built onto them cost way more than boards with just IDE, which is something I can't understand. The controller chips can't be very expensive anymore (if you use Moore's law) but I figure the companies keep the prices high because they are figuring they will sell to corporate customers with oodles of cash. I personally think the standard should be changed from IDE and SCSI to IEEE 1394. My reasoning is this, since all 1394 devices can work independantly from the CPU and from each other it would make the system all around more efficient if your DVD, HDD, scanner, and video camera were all connected to the same bus. The speed isn't bad either, 400 or 800 Mbps which is comparable to ATA-66 and faster SCSI speeds. I also rather like the idea of power being supplied by the bus itself rather than by a separate connector, that leaves me with alot less wires hanging around in my system.
I'm a loner Dottie, a Rebel.
STOP! Pay troll.
I don't see why they don't make SCSI versions of
the same drives they make as IDE. This would
make SCSI drives much more affordable.
We need a scsi price war...
Mark
-----
Don't call my crazy, that's what they called me back in the home!
4 drives my arse.
2 drives per channel. My mobo has four channels (8 drives). IDE does not need to be built in either. If you want more (or an IDE RAID) get a PCI controller.
Before putting that old drive into service after it has lain dormant, set it on top of an always on monitor and let it cook for a day or so, then place it on a flat surface and play spin the bottle. This will cause the platters to spin inside the housing. This will unstick the heads if the thermally caused expansion of the various metals involved hasn't already done the job. Then hook it up to a PC and run it for several hours with something like the seek test in PC Technician. Then copy off anything you need to save, low level format, media test to make sure any bad sectors are detected and marked, and see if you can get it to partition, high level format, boot and run.
I see even classic Slashdot is now pretty much unusable on dial up anymore.