To raid or not to Raid

C

Chickenmon

Weaksauce
Joined
Aug 27, 2004
Messages
127
So after reading anandtech, storagereview and max pc latest issue I find out my expsenive raid 0 array is actually nothing but a gimmick. I must say it does feel faster but the benches look like its just in my head. I have 2 wd raptor 36 gigs, 1 maxtor 250 gig 7200 8 meg and 1 80 gig special edition wd.

I would like to have all my apps and os on the 250 gig for simpliticitys sake, and all my downloads and various crap im working on on the 80 gig. So then I have 2 36 gig raptors to small to be useful to fast to be considered crap. Should I just pull these drives from my computer and give my powersupply a break or should a setup a raid1 and keep all my most important files on them?

I do realize using 1 of those raptor drives for my os would be the fastest way to go but 36 gig is just to small to be useful to me when I will have about 100 gigs of games and apps loaded.

I feel like a sucker and blame it on hardocp and maximumpc for not doing thier homework before pimping a new idea. I thought I did my homework before I made the purchase by reading up on the technology via hardocp and maxpc plus all the people in the forums proclaiming raid 0 is god.
 
RAID0 *is* fast. Whether or not *you* will gain from it is
an entirely seperate question. There are utilities that will
show you what your computer is doing. Use them, instead of
guessing.
Using the two 36g drives as RAID1 for important files is
definately a good idea.
 
It may be a gimmick that some people will tell you.

But I went from a single raptor to a raid setup with striping with 2 of them and I notice a difference in a lot of things.

Here are some examples that I have found.

Loading levels in games, Loading new zones on mmorpgs
Loading the OS
Loading all my applications, Especially photoshop

I feel the speed increase, I don't need benchmarks to show me what exactly the raid 0 only helps me in. Maybe it isnt enough for some people to justify at the expense of data reliability, but many of those people that have that problem. For how many years did they run on a single drive with no thought of raid at all.

It should go without saying, that all files that you find important should be backed up. For me the extra speed that I notice is worth it even if I have to rebuild an array every 6 months and spend 1 day reinstalling all my programs and using my backups that I have for my important information.
 
Well I did it on 250 as primary drive now and the 2 36's are in raid 1 as a backup drive. Ive noticed a speed drop when booting windows at least a good 2 to 3 seconds added on to it but everything else seems to be running as fast as ever. And what is supernova? :rolleyes:
 
takes alot longer than that for my RAID 5 to initialize at boot
dont get that supernova reference
 
i was referring to the fact that no one has 100 gigs of apps unless they download everything they can find
 
Now that I have a system with Raid 0... i would never go without it again


I noticed a big increase in speed... and I will never look back
 
MikeF98765 said:
100 gigs of apps? you need to stay away from suprnova
Click to expand...

Hahaha, I get it, even if no one else does. If they did, they would have spelled it right (suprnova) :D

I can't even imagine having 100gb of apps...i think 20gb is way too much :eek:
 
When I ran my Raptors in Raid 0 (36gb) I noticed they were marginally slower or basically the same than my other system running a single 36gb raptors..

These were tested on my 2 gaming PC's at the time (Dell 400SC with P4 3.2ghz and Generic Sil PCI Sata raid cards..)

Tried again with the 74gb Raptors and same results...there jusnt wasnt a tangible difference in speed between Raid 0 and a single Drive..(These were tested on my 2 currrent gaming PC's Abit Socket 939 KT800Pro with FX-53 chips and onboard Raid 0.)

I would launch games at the same time with one system using raid 0 and the other running a single drive. Would view real world testing.. boot times, level loads etc..

But I took the extra 74gb Raptors I had and made them into a 4 disk Raid 5 set in my main Dual Opteron workstation..

so in the end it worked out well..
 
Ice Czar said:
takes alot longer than that for my RAID 5 to initialize at boot
dont get that supernova reference
Click to expand...


what do you mean you by this "dont get that supernova reference" ? I think I actually know somethingn that Ice Czar doesn't!!! wo0t

suprnova.org
 
I've done a lot of testing with games, and so have quite a few people on here. RAID0 arrays add nothing to gaming. I once bought into this myth too that RAID0 was great. I had two 36 GB Raptors. I was thinking they were so much faster. In reality it's not. I tested often to find that my seek times actually went up with RAID. My games sure as hell didn't load any faster. I broke the array, started using 1 single Raptor and have suffered zero performance loss.

You can argue as much as you want that it is faster, but more and more respectable sites are proving it's not. It was a tough pill to swallow, but you can't argue with the facts. You don't gain anything. Add this to the fact your MTBF is cut in half, and it's very hard to justify RAID0 anymore.
 
USMC2Hard4U said:
Now that I have a system with Raid 0... i would never go without it again


I noticed a big increase in speed... and I will never look back
Click to expand...

What do you use your computer for? Where do you see a performance
increase?
 
MaMMa said:
suprnova.org
Click to expand...

thankyou for wising me up
however we will now promptly forget we ever had this conversation and never mention it again :p

seriously
that is a little too far into the grey area to discuss, allude to or link ;)

I remember my first mod warning
(and the only one if you discount Admin brow beating of new mods :p )
I linked a very niffty networking tool
but the site it was on, had other not so benign tools
and direct links to some very very bad sites :eek:

we try to avoid even <cough> alluding to illegal activities if you get my drift
"Follow me. Follow me. That's good, that's good! A nod's as good as a wink to a blind bat! saynomore saynomore nudge nudge wink wink"
valid bittorrent distros are generally linked at the homepage\source
 
shieldforyoureyes said:
What do you use your computer for? Where do you see a performance
increase?
Click to expand...

I do everything on my computer. I make / rip music, same with DVD's and home movies, I game, I use it as a personal office, I do it all... and I am on my computer about 6 hours a day... most of the time multitasking... thats why i got raid and p4 with HT ... I almost i wish i would have gone scsi...
 
USMC2Hard4U said:
I do everything on my computer. I make / rip music, same with DVD's and home movies, I game, I use it as a personal office, I do it all... and I am on my computer about 6 hours a day... most of the time multitasking... thats why i got raid and p4 with HT ... I almost i wish i would have gone scsi...
Click to expand...

None of those uses are in the very very tiny category of "Benefits from Raid0". I could sit and show you articles and facts all day to prove RAID0 doesn't help in any of those uses, but I'm not worried about changing people's minds....you can make up your own. I will give you some advice though. If your going to go against convention, and keep your RAID0 array....you MUST MUST backup your data an continue to keep good backups. For me personally, I can't justify running RAID0 at all....I can't find a reason that negates the risk of data loss.
 
Ice Czar said:
potentially flawed articles, benchmarks and traces
http://hardforum.com/showthread.php?t=806689
not quite as black and white as once suggested
Im still digesting that and awaiting further developments (read usage patterns)
Click to expand...

I can tell you through my own testing with a stop watch, using Norton Ghost to keep the images clean, and I could not get any improvement at all loading various levels of various games between a RAID0 array of 2 Raptors, or a single one.
 
The thing that keeps coming up is that whenever people start saying RAID 0 has little or no benefit, they are talking about fast drives like the Raptors. I agree when refering RAID 0's no performance increase when using Raptors, but take a pair of slower drives and the story is different. I have an old pair of 20GB 2MB 7.2K drives. I put them on my AMI MegeRAID 100 controller and set the stripe size just right, and I get huge increases in performance. Almost double, but that was with two slow drives that had performance in the 20MB/sec area. I would get almost 39MB/sec out of the RAID array, and any games such as SOF2 would dramaticly load faster than when on a single drive. But with the spead these Raptors have, other factors are now showing up keeping the array from scaling up like it should in terms of real world performance.

Just my two cents. I definately am not going to waste my time on putting modern drives on a RAID 0 array. No perfromance increase and now that everyone knows that it is bogus, the "bragging" factor makes the person look like a noob.
 
Met-AL said:
The thing that keeps coming up is that whenever people start saying RAID 0 has little or no benefit, they are talking about fast drives like the Raptors. I agree when refering RAID 0's no performance increase when using Raptors, but take a pair of slower drives and the story is different. I have an old pair of 20GB 2MB 7.2K drives. I put them on my AMI MegeRAID 100 controller and set the stripe size just right, and I get huge increases in performance. Almost double, but that was with two slow drives that had performance in the 20MB/sec area. I would get almost 39MB/sec out of the RAID array, and any games such as SOF2 would dramaticly load faster than when on a single drive. But with the spead these Raptors have, other factors are now showing up keeping the array from scaling up like it should in terms of real world performance.

Just my two cents.
Click to expand...

RAID arrays are still dependent on the drives that make it up. So, if your taking two slow drives and RAID0, the same findings will apply when you compare one or the other. I had some synthietic benchmarks to show my arrays giving me huge performance differences. It's just not a real world test. It's the same deal as video benchmarks. This debate came up about a month ago, so I repeated my tests using two Maxtor 20 GB 7200 rpm drives and found the same thing. Synthetic tests showed a big difference, but real world testing showed none.
 
To anyone who thinks that their RAID0 array "Feels" faster - perhaps a stopwatch? I'll admit that it' is a pain in the ass to rebuild your system jut to prove that you actually get a speed increase - but attempting to argue against facts with opinons is spurious at best.
 
[H]Rabbit said:
To anyone who thinks that their RAID0 array "Feels" faster - perhaps a stopwatch? I'll admit that it' is a pain in the ass to rebuild your system jut to prove that you actually get a speed increase - but attempting to argue against facts with opinons is spurious at best.
Click to expand...

I agree with you, in your points. But, if your using Norton Ghost Corporate with some clean images, it gets very easy to go back and forth.
 
djnes said:
RAID arrays are still dependent on the drives that make it up. So, if your taking two slow drives and RAID0, the same findings will apply when you compare one or the other. I had some synthietic benchmarks to show my arrays giving me huge performance differences. It's just not a real world test. It's the same deal as video benchmarks. This debate came up about a month ago, so I repeated my tests using two Maxtor 20 GB 7200 rpm drives and found the same thing. Synthetic tests showed a big difference, but real world testing showed none.
Click to expand...
I am not pulling this shit out my ass here djnes. There was a real world improvement. I did not do the stop watch, but it was more than just feel. Loading a new level in SOF2, I went from last in to first in. Take into consideration that this is only under 40MB per second on the PCI bus, vs theoretical 100MB per second (just a guess) with two Raptors. Someday when I get time, I will dig out my old IWILL KK-266-R and set this up again and find out how much of an actual improvement there was.

I think there is a point at which the speed of the data coming off the drives doesn't matter anymore. It seems like the CPU is busy decompressing the pak files and such and other overhead. That's why I think I saw an improvement going from 20MB/sec to 39MB/sec, whereas going from 50MB/sec to 100MB/sec shows less of an improvement...percentage wise.

Don't get me wrong, I am not argueing wether or not StorageReview and Anandtech were right or wrong. I have 100% faith in the fact that they were right. I just think that you take a pair of slow drives from circa 1999/2000 and pair them up in RAID0, your gonna get some benefits..albeit, they are probably not going to scale anywhere near 100% in real world performance.

During your test did you try different stripe sizes? I found that going too large yeilded no improvement in real world if the read/writing was smaller than the stripe, since it used only one drive, where as too small of a stripe, used too much CPU overhead. Finding the balance is important. If my memory serves me correct, I was using a stripe of 64kb.
 
Most people just tested RAID with synthetic benchmarks at first, everyone figured if it was server technology it must make your desktop faster (obviously not, in most cases)... Anyway, having a RAID 1 array made up of two expensive 36gb Raptors is kind of a waste of them (and money), particulary if you're not actually running anything off them directly.

You don't need the speed for backup purposes and the size doesn't help even if you have less than 36gb of photos, docs, music, etc. Actually, I don't see how you think you could fit all your "most important files" within 36gb yet you somehow have 100gb of just apps and games?

The OS itself and basic essential apps doesn't take much more than 6gb... You either have way too many apps that you probably don't use or that's a lot of monster-install-size games you've got loaded (that you probably don't play either :p ).

In any case, I'd either sell both Raptors or use one of them for the OS and most essential apps while keeping the rest of your stuff on the 250gb drive, then store documents or backup the OS 36gb driveon the 80gb one. Most effective solution imo, you're still left with one extra Raptor tho... Could just RAID1 it and not have to backup the main OS drive ever I guess, the performance hit won't be much (probably be faster than using the 250gb as your OS drive still).

I still don't get how you manage 100gb of games and apps...
 
P.S. Anyone try Ghost 9.0 yet? Is it supporting SATA drives correctly? I read an article that said it would at release but it still wasn't when tested, and do older versions perform scheduled backups to external drives? 9.0 won't apparently...

I've traditionally done backups of data only but I wanna start imaging (seems like the easiest way of doing full sys backups) my system on a regular basis to be ready in the event of another drive failure (first an IBM Deathstar two years ago and now an old 40gb Quantum drive I was using for music), trying to figure out the best method.
 
Met-AL said:
I think there is a point at which the speed of the data coming off the drives doesn't matter anymore. It seems like the CPU is busy decompressing the pak files and such and other overhead. That's why I think I saw an improvement going from 20MB/sec to 39MB/sec, whereas going from 50MB/sec to 100MB/sec shows less of an improvement...percentage wise.
Click to expand...
From another thread (http://hardforum.com/showthread.php?p=1026646946#post1026646946):

http://www.tweakers.net/reviews/515 said:
Our gaming traces show that most games have very little impact on modern hard drives and generate a rather low load. Even the heavyweight champion Battlefield Vietnam was able to burden a Raptor WD360GD for only 17,1 percent average with a short peak of 70 percent. Other games had even lower averages and didn't rise above 80 percent peak usage. Therefore we conclude that loading game levels is mostly cpu intensive and does not rely on storage devices. This conclusion is again illustrated by the insignificant improvents when loading levels in Far Cry and Unreal Tournament. Also, measurements with Unreal Tournament show some unlikely differences between a Raptor WD740GD and the other disks in the arena. It seems to be 12,9 percent faster than number two, a Seagate Barracuda 7200.7, while the others (one of them is a Raptor 360GD and a rather old IBM Deskstar 75GXP) offer a performance varying not more than 8,4 percent.
Click to expand...
 
Impulse said:
P.S. Anyone try Ghost 9.0 yet? Is it supporting SATA drives correctly? I read an article that said it would at release but it still wasn't when tested, and do older versions perform scheduled backups to external drives? 9.0 won't apparently...

I've traditionally done backups of data only but I wanna start imaging (seems like the easiest way of doing full sys backups) my system on a regular basis to be ready in the event of another drive failure (first an IBM Deathstar two years ago and now an old 40gb Quantum drive I was using for music), trying to figure out the best method.
Click to expand...

Ghost 2003 and Ghost 9.0 work fully and completely with SATA and SATA RAID0 drives. I think it has something to do with the fact that Intel's implementation of SATA is a little more friendly and compatible outside of Windows than the AMD boards. But I've always laughed when I read an article saying they weren't compatible. They both work completely fine for me.
 
Well what I read about Ghost 9.0 and SATA was just a comment off a PC World review. Not the most reliable source, I know, was looking for some reviews on it since it hasn't been out long tho, and I know it changed some after Symantec acquired Powerquest's Drive Image.

What made me ask was I remembered having read a longer article on Maximum PC about Ghost 2003's problems with SATA drives. I digged up the article and it was kinda unconclusive, they said they could create images just fine with Drive Image 7 but they wouldn't restore right and the company didn't offer support for SATA backup. They said that to get it to work with Ghost 2003 they had to use a PCI controller card (wouldn't work on a 865GFB board)... Oddly enough they reported no problems with the even older version of Drive Image (2002), kinda puzzling.

Recently (last issue) someone wrote in saying none of the DI versions nor Ghost 2003 would correctly image his complex Adaptec-driven RAID arrays but that Arconis True Image 7 would. Despite all this the editor still pointed out in his response that Symantec stated officially on their site Ghost 2003 supported SATA.

Not really sure what to make of it all, sounds like some of the problems may indeed stem from RAID arrays or particular controllers, nVidia's latest SATA implementation is pretty hassle-free as well. I'll likely pick it up and try it out in any case, hope it does allow me to schedule backups to external drives or at 'least move them there once done (rather use my spare intenral on another rig). Thanks for the info tho, 'least I know it works in some configs, hopefuly kinks have been worked out since the 2003 version.
 
nothing affordable officially supports RAID
not that some dont manage it anyway

but from syamantec
http://service1.symantec.com/SUPPOR...003 for Windows 2000/NT/Me/98/XP&osv=&osv_lvl
regardless of whether the drive uses software level RAID or hardware level RAID. Successfully imaging a RAID drive is dependent on the specific computer model, driver controller, hard drive, and RAID implementation.
Click to expand...

simple levels are generally successful, but parity levels rarely if ever are
without some serious $$$$

that pretty much holds true across most imaging products
simply backing up RAID arrays is generally perferable to imaging them
 
I was more concerned with simple SATA support than RAID support (I'm not running any sort of RAID array atm myself), it seemed these programs were having trouble even with that (single volume SATA drives, period)...
 
Impulse said:
I was more concerned with simple SATA support than RAID support (I'm not running any sort of RAID array atm myself), it seemed these programs were having trouble even with that (single volume SATA drives, period)...
Click to expand...

If your running an Intel chipset board, I can tell you the last two consumer versions of Ghost will work for sure.
 
Ghost 2003 did problems with 865 chipsets. It's a known issue and they've fixed it. Run liveupdate and it'll fix you up. Btw, It's not just SATA either; you'll have problems ghosting even IDE drives without the patch(which is where I ran into the issue).
 
the raid 0 findings going on now have been known by those of us who use raid a lot for a long time.

If all you do is game then Raid 0 will be of little to no benefit to you. Raid 0 did provide massive increases whenn gaming back when vid cards had very little ram and used the swap file a lot. Any data intensive program that constantly uses your swap file will benefit from raid 0.
 
Well... :D

Think of it this way. By raid 0 - ing your two 36 gb drives anyway you do end up with 1 70gb drive that everyone here seems to agree is at least not going to be slower than the 36's separately. You will have one nice fast 70ish GB drive. That won't cut it for your 100MB os/apps, I realize, but it is something to consider.
 
tcraig_4096 said:
Well... :D

Think of it this way. By raid 0 - ing your two 36 gb drives anyway you do end up with 1 70gb drive that everyone here seems to agree is at least not going to be slower than the 36's separately. You will have one nice fast 70ish GB drive. That won't cut it for your 100MB os/apps, I realize, but it is something to consider.
Click to expand...


OO OOO OOO!!!! and when one of the drives fails I can lose ALL MY DATA!!! ;)
If you really want to combine the drives you can always use windows dynamic disks to mount one in an empty folder on the other.
 
Eh, tru but I always save my few megs of important stuff on cd's or another hard drive. As someone else said, it is no different from the old days when we had one hard drive and no back up. I was just giving a suggestion for a good use of his (still very nice and fast) drives.

Nothing says that the probability of a drive failing increases just because you have more of them.
 
tcraig_4096 said:
Eh, tru but I always save my few megs of important stuff on cd's or another hard drive. As someone else said, it is no different from the old days when we had one hard drive and no back up. I was just giving a suggestion for a good use of his (still very nice and fast) drives.

Nothing says that the probability of a drive failing increases just because you have more of them.
Click to expand...

Wait a tic... you're telling me that a computer with 1 hard drive in it has the same MTF as, oh, say an entire server room at a graphics company? (They generate several hundered Gig a DAY). If that is your interpretation of the laws of probability, then you, my friend, are an idiot. Congrats!
 
You must log in or register to reply here.
Back
Top