Raptor Hds?

[phaelinx]

Ok I have a ata133 250gig maxtor hd, and 2 western digital 120gig ata100 hds. Currently I use the 250gigger for windows and the 2 120's in a raid0 setup for storage. If I add one of those 36gig Raptor 10k rpm hds, and put windows on that how much of a performance increase in load times for windows/games would I see? And how much of a performance increase would I see if I bought 2 of them and put them in a raid0 setup?

[MaMMa]

thats basically the same system setup that I have. I just got my raptor in 2 days ago. I notice bootup so much faster and windows apps are very much faster. Everything feels a lot better. You shouldnt' regret.

[tdg]

Whatever you do, get your data off that RAID 0 setup, and onto a single drive or RAID 1 if possible, RAID 0 is just asking for data loss.

[GJSNeptune]

What are the differences between RAID 0 and RAID 1, anyway?

[phaelinx]

so what do you recommend i have the 2 hds at? raid1? and raid0 causes alot of data loss??

[MaMMa]

correct me if im wrong.

raid 0 if one drive crashes then u lose both. i read it somewhere in this forum " would you turn in a paper that has every other word " (something like that)

raid 1 i think one drive is still good?

[phaelinx]

well my motherboard comes with a raid managing program that allows me to unlink the 2 drives even if 1 fails. im not worried about it failing, im worried about the data..but thats another story, im just interrested in a performance increase on my system..

[tdg]

RAID 0 is where two drives act as one. One chunk of a file is written to one drive, the next chunk goes on the other drive and back and forth. When one drive fails you loose the data on both, and since the chances of a drive failure is higher with using more drives, its asking for trouble.

RAID 1 is when both drives have the exact same data copied onto them. If one drive dies the other drive still has all data intact. The downside to RAID 1 is that its inefficient for storage, since if you had 2x 200 GB drives, you only have 200 GB usable since both drives have the same data.

[phaelinx]

so what do you recommend i do with my 2 120 giggers..? I want to raid them to gain performance but not at the cost of failure..i do however want the space of both hds..if i abolutely have to i'll raid1 them..but that would suck to just pretty much lose 120gigs of space....

[aug1516]

First off, what kind of things are you doing with this 240gb raid array right now? Do those activies even take advantage of the array's abilities? If not, then you can take them out of an array and just use them seperately. That would still leave you with half your data if a drive fails and you would not lose any space. What kind of backups do you run on this array if any at all? If you run frequent backups then you can really do whatever you want with them because you won't lose much data. If backups are less frequent then Raid-1 might be the answer depending on the importance of the data.

[Leadman584]

RAID0 should not be considered a storage option, under any circumstances. RAID1 possibly but you lose half available disk space, since half the array is pure back up, identical copy of the other half. RAID5 or 50 is how corporations do it, but they also use large number of drives. I can do it with 4 drives on one card, and 5 drives on the other. Performance using even small raptors exceeds, any ATA solution or even a single large raptor. If pretty good performance without excessive cost is the goal here, a single 74GB raptor for OS and games is probably the cheapest solution. If you do huge sequential reads, writes (video editing), RAID5 should be you're first choice, purely for the sake of safety. One drive will do nothing but run parity data to ensure the rest don't become corrupted beyond use, and if a drive fails, will rebuild all data onto remaining drives.

[defakto]

well my motherboard comes with a raid managing program that allows me to unlink the 2 drives even if 1 fails. im not worried about it failing, im worried about the data..but thats another story, im just interrested in a performance increase on my system..


If it's raid 0, even if you can "unlink" the drives you still lose EVERYTHING on the array.

[phaelinx]

I already moved all the data to my 250 gig hd, and unlinked the 2 120s and now have them as separate drives..They are storage so i didnt want corruption..so you need 4 drives for a raid5?

[TheGameguru]

well my motherboard comes with a raid managing program that allows me to unlink the 2 drives even if 1 fails. im not worried about it failing, im worried about the data..but thats another story, im just interrested in a performance increase on my system..

you need to re-read the definition of Raid 0 to understand why that wouldnt work.

[aug1516]

I already moved all the data to my 250 gig hd, and unlinked the 2 120s and now have them as separate drives..They are storage so i didnt want corruption..so you need 4 drives for a raid5?

You would also need some sort of Raid card that supports Raid-5 unless you want to use Windows build in software support. I believe you need a server package for that feature though. Raid-5 is one of the best ways to go but it's just not feasable for a lot of normal users that can't afford expensive hardware Raid-5 cards. Keep them seperate and just make good backups of the data. If you need to buy a DVD burner to do that then go get one.

[phaelinx]

you need to re-read the definition of Raid 0 to understand why that wouldnt work.

what do you mean wouldnt work? I just wanted the benefits of Raid..I am not very familiar with it. So I did setup a raid0 for awhile, but I moved all the data off the raid drives, deleted the array setup, and reformatted both hds in windows disk management. Now both drives are separated and I just moved my data back onto the drives..I'm just tired of the windows swap file being a bottleneck on my system because my drives are so slow..I mean its not that slow, but exiting games, my hds spins for about 5 seconds causing windows to load the desktop slowly

[eastvillager]

Raid 0 with two drives is perfectly fine for applications and temp data. It is perfectly fine for all your data, if you maintain reliable backups. This is talking about home use, where downtime due to hardware failure is acceptable. Work ends up being mirrored stripes, or RAID 5.

I backup a 2x74 raptor raid 0 to a cheap 180 gig drive every couple days, and have no fear of losing one of the raptors.

Anything that qualifies as actual data I've created ends up on removable at some point as well.

[eastvillager]

He means it wouldn't work AFTER a drive failure, as you'd have just lost half your data.

What you probably mean is that your software will deconstruct a striped pair for you. It of course would have to be a functional stripe, though, not one where a drive had failed.


what do you mean wouldnt work? I just wanted the benefits of Raid..I am not very familiar with it. So I did setup a raid0 for awhile, but I moved all the data off the raid drives, deleted the array setup, and reformatted both hds in windows disk management. Now both drives are separated and I just moved my data back onto the drives..I'm just tired of the windows swap file being a bottleneck on my system because my drives are so slow..I mean its not that slow, but exiting games, my hds spins for about 5 seconds causing windows to load the desktop slowly

[phaelinx]

He means it wouldn't work AFTER a drive failure, as you'd have just lost half your data.

What you probably mean is that your software will deconstruct a striped pair for you. It of course would have to be a functional stripe, though, not one where a drive had failed.

well I load the Raidmanaging software up and it shows both hds linked into 1 drive, then the options below it say "Create Array, Delete Array, and Rebuild Array" I assumed that since its showing both drives and the array that even after a drive failure i would still be able to delete the array..correct me if im wrong, but thats what it looks like..

[djnes]

If you think the page file is the bottleneck of your system...why use one?

[phaelinx]

If you think the page file is the bottleneck of your system...why use one?

well with all the previous versions of windows, isnt it "recommended" to have one even if you have more than enough ram? I guess I could disable it but would it cause random windows errors?

[liquidtrance123]

i love my Raid 0 Raptors they're :eek: :D :cool:

[djnes]

well with all the previous versions of windows, isnt it "recommended" to have one even if you have more than enough ram? I guess I could disable it but would it cause random windows errors?

Your never going to find a definitive yes or no. However, many many people do not have a page file, and have no problems. I think it's one tweak where you can feel the difference. Every PC I ever build feels snappier without one...assuming it has enough memory. Three of my systems have 1 GB of memory, and none of them have a page file.

[djnes]

i love my Raid 0 Raptors they're :eek: :D :cool:

Yikes...please tell me you don't still subscribe to that myth!

[phaelinx]

Your never going to find a definitive yes or no. However, many many people do not have a page file, and have no problems. I think it's one tweak where you can feel the difference. Every PC I ever build feels snappier without one...assuming it has enough memory. Three of my systems have 1 GB of memory, and none of them have a page file.


Ok, i have a 256 meg ram drive setup, would be good to setup a 200 meg pagefile on that drive or just remove it all together?

[djnes]

Ok, i have a 256 meg ram drive setup, would be good to setup a 200 meg pagefile on that drive or just remove it all together?

200 MB doesn't seem worth the trouble at all anyway. I don't really know your uses of the computer, and what software you run...but it can't hurt to try it without a page file. If you start getting out of memory errors, or apps complain about no virtual memory, then you'll have to re-add it. If I ever have a page file, I set it as a static size of 400 MB and I just leave it on the C: drive.

[liquidtrance123]

Yikes...please tell me you don't still subscribe to that myth!
:confused: wtf are you talking about?

[djnes]

:confused: wtf are you talking about?


WTF am I talking about? The fact that it has been proven and beaten to death that RAID0 does absolutely nothing for you in a desktop setting, besides cutting your MTBF in half. I'd suggest some heavy reading at Anandtech and StorageReview before you go around touting RAID0.

[MaMMa]

lets not kill this thread guys

[phaelinx]

I use my pc for web browsing, chatting, and gaming, and posting on [h] baybee!...anyways..um...im just tired of exiting a game ive been playing for a couple hours to have windows/my hd lag as it attempts to reload the desktop icons and background/screen resolution, etc..shits annoying..i mean with my current setup, gig of ram, 3200+ you'd think it wouldnt do that but it does...i think its either 1. my hds or 2. a lousy vm/swap file thats slowing it down

[v3rt1g0]

djnes, don't take a steaming dump on this thread.

I've said it before, and I'll say it again, you're bashing something you don't or can't have to feel better about your single disk's performance. Unless you have something useful to add, please don't 'contribute'.

--

I'm going to ignore the anandtech topic even though I've got plenty to say about it.

Personally, I have two 36.7g raptors in RAID-0, and I backup the array to a single 80GB removable PATA drive once a week. No worries here.

The speed increase in absolutely noticeable. I went from a PCI SATA Raid card to my current motherboard with onchip RAID (bypasses the PCI bus), the the speed increase was stunning. The raptors were being bottlenecked by the PCI bus before.

phaelinx, a question. Do you have onboard SATA RAID that bypasses the PCI bus? Like some (all?) of the nForce3 chipsets and kinda recent P4 chipsets. If you have an older motherboard, the raptors wouldn't be used to their full potential.
If you do have a suitable motherboard, I'd get two raptors, load windows and all your games and any other high i/o apps (photoshop scratch), and then REGULARLY backup the raptor array to your 250gb drive with norton ghost (creates an EXACT replica. very very easy system restores).

EDIT: Didn't see your sig while I was in the message editor. I JUST upgraded from the mobo you have right now. I'm sorry to say, but the onboard SATA controller IS THE DEVIL. The drivers for it just suck. BUT, see if they've been updated, they could have fixed them. They would bluescreen windows (both 2000 and XP) relentlessly. It was a known problem back then.

I THINK, the Silicon Image 3112a SATA controller runs across the PCI bus, someone correct if I'm wrong here. Therefore, I'd recommend a single raptor drive for now. Get another one later on once you upgrade your motherboard.

[dandragonrage]

Never put a pagefile on a RAM drive. Instead, don't use a RAM drive. If you did put a page file on a RAM drive, it's going through the paging subsystem then through the RAM drive subsystem to accomplish what it could otherwise do going straight to RAM... Bad stuff.

[phaelinx]

vert1go, I have 2 onboard raid controllers, one is for my ide3 and ide4 ports..its ran by an ITE gigabyte driver and the other is a SataRAID controller with sata1 and sata2 ports..both controllers bypass the pci bus..and ive taken your guys' advice and didnt put the swap file on the ram drive..instead i disabled it..and am now just putting my temp folders/temp internet files on the ram drive so taht when i reboot...it auto deletes it..which is nice...just with doing those 2 things i noticed a huge increase in load times and less hd spins..is that because the page file isnt being resized dynamically while playing games/loading stuff ?? or is it more because i moved a bunch of small temp files off the drive and into the ram drive??

[v3rt1g0]

vert1go, I have 2 onboard raid controllers, one is for my ide3 and ide4 ports..its ran by an ITE gigabyte driver and the other is a SataRAID controller with sata1 and sata2 ports..both controllers bypass the pci bus..

You positive about that? Let me do some looking..

I'm aware of the PATA ITE controller. Does the ITE chip still perform really badly in RAID 0? Last I saw, single drives were outperforming RAID 0 arrays using the ITE controller. Bad implementation on Gigabyte's part I'm told.

[v3rt1g0]

Found it.

http://www.siliconimage.com/products/sii3112.asp

"Featuring a 32-bit 33/66 MHz PCI interface, the SiI 3112 supports two independent Serial ATA devices and can be incorporated in a motherboard design or a PCI add-in card"

Meaning, even though it's not a PCI *card*, it still runs it's data through the PCI bus.

[v3rt1g0]

... is that because the page file isnt being resized dynamically while playing games/loading stuff ??

That'd be my conclusion. Glad to hear your system is running faster :)

[phaelinx]

Yeah the performance is still slow...I had 2 ata100 7200 hds raid0 and were performing slower than the raid0 ata100 5400 hds on the same benchmarks..im wondering if thats just a driver issue or if its an actual hardware issue? either way once generation 2 pcix boards along with gen 2 amd64 bit boards (plus 64bit winxp) is out then i'll be upgrading..as it is..i got a 3200+ at 2.2ghz but my mb says it supports up to 3.2ghz...so i might screw around with overclocking it..cause someone mentioned 3200+ is probably the last line of xp processors...which blows...why put support for a 3.2ghz processor on a mb if you cant even hit that mark..anyways...i'll have to try and oc it some..

[liquidtrance123]

WTF am I talking about? The fact that it has been proven and beaten to death that RAID0 does absolutely nothing for you in a desktop setting, besides cutting your MTBF in half. I'd suggest some heavy reading at Anandtech and StorageReview before you go around touting RAID0.
i noticed quite a large difference between 1 raptor and 2 in raid0
besides i work with quite large files all the time between my main PC and my Raid5 array on my webserver. files are at least 100mb and up to 700mb per file. So i believe i am getting much better performance from my array than with only 1 drive.

edit: and being as both my drives come with 5 year warranties i'm far from worried about my drives failing.

[stryder2720]

Why not just but one 74gig Raptor?

[djnes]

edit: and being as both my drives come with 5 year warranties i'm far from worried about my drives failing.

The warranties don't cover lost data....and the minimal difference in performance is not worth the risk of data security. I work with large files in the form of video, and no real world benchmark tests showed any difference in performance worth considering. As I mentioned before...read Anand's article.

[liquidtrance123]

The warranties don't cover lost data....and the minimal difference in performance is not worth the risk of data security. I work with large files in the form of video, and no real world benchmark tests showed any difference in performance worth considering. As I mentioned before...read Anand's article.
well i'm glad my drives being in raid 0 seem to piss you off so much but they're staying. And it won't matter if my drives die all my important data is on CDR, my external 120 (which is usually tucked away till i need it) and my Raid5 array downstairs.

so if my Raid 0 goes down i just have to reload my OS and thats all.

its kuz of posts like yours that i stopped coming here so often.

[phaelinx]

how much of an increase if any does raid0 offer? or is it depedant on drive specs etc?

[dawtips]

well i'm glad my drives being in raid 0 seem to piss you off so much but they're staying. And it won't matter if my drives die all my important data is on CDR, my external 120 (which is usually tucked away till i need it) and my Raid5 array downstairs.

so if my Raid 0 goes down i just have to reload my OS and thats all.

its kuz of posts like yours that i stopped coming here so often.

Ya I pretty much have the same setup and I certainly notice a difference with raid 0 74gig raptors.

[liquidtrance123]

Ya I pretty much have the same setup and I certainly notice a difference with raid 0 74gig raptors.
Apparently according to djnes Raid0 is satan and completely worthless so you should break up your raid array now and jump to raid 1 ASAP as to please his greatness. :rolleyes:


i love my raid 0 much faster than just 1 drive.

[liquidtrance123]

how much of an increase if any does raid0 offer? or is it depedant on drive specs etc?
at certain things Raid 0 is fairly nice, however if your just doing games and word processing its probably not for you as you won't notice "MUCH" difference.

however as djnes so eloquently explained to me Raid 0 sucks. So i guess he's right and we should all break up our raid 0 arrays.

[TheGameguru]

I couldnt care a less if you keep or break your Raid 0.

Just please dont continue to spread the "myth" of its huge performance difference when frankly there is none what so ever unless your peforming very specific tasks.

For general Windows and Gaming its actually worse than running the same drives in single drive mode.

I run my 4 74gb Raptors in Raid 5 with an Adaptec SATA Raid Card and its noticably slower than my other systems running a single 74gb Raptor but on my main system I care more about my data than my gaming systems.

Raid is for redudancy and protection.

[djnes]

Just please dont continue to spread the "myth" of its huge performance difference when frankly there is none what so ever unless your peforming very specific tasks.

Thank you very much.

LiquidTrance...if you have something to say to me...PM me. Don't act like a little kid with the sarcasm. No one finds it funny. Your going against accepted fact. Don't argue with me on this. E-mail Anand from Anandtech, and tell him his article is wrong. Again...I'll say it once more. Your childish bullshit probably won't be tlerated anymore in the public forums. If you have more ridiculousness to spit out of your mouth through your keyboard, do it through PM.

For those who want actual proof, here is the article...linked right to the conclusion.
http://anandtech.com/storage/showdoc.aspx?i=2101&p=11

[GreNME]

The Anand article does a great job of showing the results where a RAID setup is not offering any speed increase. That it's being constantly ignored by people as an indication that the claims to the contrary are based on myth is astounding. However, allow me to give some brief explanation...

RAID was never meant to give anything any more speed, because that's not the point of what RAID is. RAID is an acronym for Redundant Array of Independent (or sometimes Inexpensive) Disks. The "redundant array" is the part that should really be stressed in this case, because this is exactly the reason to have RAID setups to begin with—redundancy.

One of the benefits of this redundancy in a server environment is an increased theoretical bandwidth for access at any given moment. What this means is that, for a server running RAID (0, 1, 5, etc.), more people logged onto this server will be able to access the contents of the disk at a faster rate. This allows the server to be able to handle more of a load with a higher guarantee of efficiency.

The thing is, this theoretical increase is often mistaken as applying exactly the same when used on a single (non-server) machine with a single user on a keyboard, mouse, and monitor connected directly to the box (as opposed to many network connections running terminal sessions). In fact, this is not true. In reality, disk access is going to be almost exactly the same as if there were only a single disk in the machine—this is why it's called a "redundant array" to begin with. That's the point. It is supposed to be so redundant that, even as John Smith sits at the machine itself and works on Project A, Jimmy Guy can be connected via a terminal session working on Project B (and, if he's really slick, Project C on the side), and Jane Doe can also be connected via the network and working on Project Z, with none of the three people experiencing a noticable degradation of performance while all working from the same disk at the same time in different ways.

This is what RAID is and what it does. This is what it was designed for. There is nothing in the mechanics of how RAID works that supports claims of increased disk performance in the manner that some have claimed in the thread. Djnes has already not only pointed this out, but shown an example where someone actually tested single drives next to redundant drives and saw negligible (and varied) differences. RAID does not increase your frames-per-second or web browser load times. With some specialized software (like high-end rendering, audio, and video editing software), a slight increase may be noticable, but not an extreme one. Regardless, RAID does not make your regular computing experience any faster.

[liquidtrance123]

Thank you very much.

LiquidTrance...if you have something to say to me...PM me. Don't act like a little kid with the sarcasm. No one finds it funny. Your going against accepted fact. Don't argue with me on this. E-mail Anand from Anandtech, and tell him his article is wrong. Again...I'll say it once more. Your childish bullshit probably won't be tlerated anymore in the public forums. If you have more ridiculousness to spit out of your mouth through your keyboard, do it through PM.

For those who want actual proof, here is the article...linked right to the conclusion.
http://anandtech.com/storage/showdoc.aspx?i=2101&p=11
You might wanna take a step back there, all i ever said was, "man i love my raptors in raid 0."

then all of a sudden it became a HUGE deal to you that apparently somewhere in "man i love my raptors" meant that there was a HUGE performance difference and everyone should rush out right now and go buy some to put in raid 0.

So if someone somewhere can get "HUGE PERFORMANCE DIFFERENCE" out of "man i love my raptors in raid 0" then i guess your right. :rolleyes: holy christ this place has gone to hell in a handbasket :(

[djnes]

holy christ this place has gone to hell in a handbasket :(

Only because of people like you who immediately degrade to childishness instead of having a friendly debate or conversation. Your several posts filled with sarcasm begged for a smartass return. And now that myself and GreNME proved a concept, your wanting to blow it off like nothing happened. Very very weak logic there. The problem with these boards is not me, my friend...it's you and the immaturity that follows when someone offers a differing opinion. No one is forcing you to stay here...this isn't your job. If you don't like being corrected, learn or go elsewhere.

[liquidtrance123]

Only because of people like you who immediately degrade to childishness instead of having a friendly debate or conversation. Your several posts filled with sarcasm begged for a smartass return. And now that myself and GreNME proved a concept, your wanting to blow it off like nothing happened. Very very weak logic there. The problem with these boards is not me, my friend...it's you and the immaturity that follows when someone offers a differing opinion. No one is forcing you to stay here...this isn't your job. If you don't like being corrected, learn or go elsewhere.
you forget this was after you automatically assumed from my simple post that i'm completely wrong and need to be immediately corrected for my "opinion" :rolleyes:

sorry if my "opinion" about raid 0 is wrong i'll immediately correct my "opinion" to match your logic

[liquidtrance123]

Ya I pretty much have the same setup and I certainly notice a difference with raid 0 74gig raptors.
apparently we're wrong :(

[djnes]

you forget this was after you automatically assumed from my simple post that i'm completely wrong and need to be immediately corrected for my "opinion" :rolleyes:

sorry if my "opinion" about raid 0 is wrong i'll immediately correct my "opinion" to match your logic

Opinions don't hold any water compared to tested fact. PM GreNME about that one. Any difference you "notice" is psychological. Now...do you want to continue the childish games, are do you want to admit you bought the myth we all did at one time, and have since learned the truth? It's your call, but I can assure yuo the mods aren't going to be happy with your games and sarcasm.

[liquidtrance123]

Opinions don't hold any water compared to tested fact. PM GreNME about that one. Any difference you "notice" is psychological. Now...do you want to continue the childish games, are do you want to admit you bought the myth we all did at one time, and have since learned the truth? It's your call, but I can assure yuo the mods aren't going to be happy with your games and sarcasm.
i'm sorry but no i'm not going to say i didn't notice a difference in speed because i did, if you don't want to believe me you go right ahead but just kuz you posted a link doesn' t mean i was "dreaming" when i noticed a difference.

[WaterIsTasty]

Why not just run the 2x120 out of raid? its not like you need lots of speed for file storage anyways...

[phaelinx]

Why not just run the 2x120 out of raid? its not like you need lots of speed for file storage anyways...


I currently am, after reading some of the colorful posts, I broke my raid0 array and now have both 120's as separate drives.. =/

[shieldforyoureyes]

The Anand article does a great job of showing the results where a RAID setup is not offering any speed increase. That it's being constantly ignored by people as an indication that the claims to the contrary are based on myth is astounding. However, allow me to give some brief explanation...

RAID was never meant to give anything any more speed, because that's not the point of what RAID is. RAID is an acronym for Redundant Array of Independent (or sometimes Inexpensive) Disks. The "redundant array" is the part that should really be stressed in this case, because this is exactly the reason to have RAID setups to begin withredundancy.

One of the benefits of this redundancy in a server environment is an increased theoretical bandwidth for access at any given moment. What this means is that, for a server running RAID (0, 1, 5, etc.), more people logged onto this server will be able to access the contents of the disk at a faster rate. This allows the server to be able to handle more of a load with a higher guarantee of efficiency.

The thing is, this theoretical increase is often mistaken as applying exactly the same when used on a single (non-server) machine with a single user on a keyboard, mouse, and monitor connected directly to the box (as opposed to many network connections running terminal sessions). In fact, this is not true. In reality, disk access is going to be almost exactly the same as if there were only a single disk in the machinethis is why it's called a "redundant array" to begin with. That's the point. It is supposed to be so redundant that, even as John Smith sits at the machine itself and works on Project A, Jimmy Guy can be connected via a terminal session working on Project B (and, if he's really slick, Project C on the side), and Jane Doe can also be connected via the network and working on Project Z, with none of the three people experiencing a noticable degradation of performance while all working from the same disk at the same time in different ways.

This is what RAID is and what it does. This is what it was designed for. There is nothing in the mechanics of how RAID works that supports claims of increased disk performance in the manner that some have claimed in the thread. Djnes has already not only pointed this out, but shown an example where someone actually tested single drives next to redundant drives and saw negligible (and varied) differences. RAID does not increase your frames-per-second or web browser load times. With some specialized software (like high-end rendering, audio, and video editing software), a slight increase may be noticable, but not an extreme one. Regardless, RAID does not make your regular computing experience any faster.

Ah.... I'm sorry, but you have no idea what you're talking about.
There's nothing about RAID that's better for multi-user loads
than single-user. The "Redundency" in "RAID" has to do with
fault-tolerance, not being able to handle a bigger load, or
"efficiency", whatever it is you mean by that.
RAID was always intended for speed as well as reliability, and
the "I" has always stood for "Inexpensive", because it was
designed to replace minicomputer/mainframe disks which were
obscenely expensive.
This is the paper that invented the concept, came up with the
term "RAID" and defined the basic RAID levels:
http://www-2.cs.cmu.edu/~garth/RAIDpaper/Patterson88.pdf
(Patterson is more well-known for being the researcher behind
the first RISC microprocessor, which was commercialized as
the SPARC architecture.)
Now, it may well be true that:
a. Most desktop users these days are not disk-io-bound, so
speeding up their disk system won't actually help anything.
b. The IO capability of PCs rather sucks, so most PCs are
probably maxed out just trying to deal with a single new
fast drive, and can't cope with a RAID array.
But neither of those things is a criticism of RAID itself.

[GreNME]

Ah.... I'm sorry, but you have no idea what you're talking about.
[edited]

There's nothing about RAID that's better for multi-user loads
than single-user. The "Redundency" in "RAID" has to do with
fault-tolerance, not being able to handle a bigger load, or
"efficiency", whatever it is you mean by that.
It's all about the I/O, which makes it optimal for multi-use situations, but the efficiency is in the I/O (and redundancy checking). In fact, efficiency is mentioned numerous times in your own link.

RAID was always intended for speed as well as reliability, and
the "I" has always stood for "Inexpensive", because it was
designed to replace minicomputer/mainframe disks which were
obscenely expensive.
It was not designed for replacing the single disk drive in terms of speed, it was designed for matching huge mainframes which are, oddly enough, multi-user environments. How interesting you don't know that about mainframes!

This is the paper that invented the concept, came up with the
term "RAID" and defined the basic RAID levels:
http://www-2.cs.cmu.edu/~garth/RAIDpaper/Patterson88.pdf
(Patterson is more well-known for being the researcher behind
the first RISC microprocessor, which was commercialized as
the SPARC architecture.)
Nothing in the paper, had you read it, contradicts what I said.

Now, it may well be true that:
a. Most desktop users these days are not disk-io-bound, so
speeding up their disk system won't actually help anything.
b. The IO capability of PCs rather sucks, so most PCs are
probably maxed out just trying to deal with a single new
fast drive, and can't cope with a RAID array.
But neither of those things is a criticism of RAID itself.
You don't get it, do you? Neither djnes nor myself have criticised RAID at all. We simply stated that suggesting it for speed—as in playing games, surfing the web, or any other sundry common task for a personal computer—is not based in fact. As I also stated in my post—had you been so kind as to have actually read it—that there do indeed exist cases where a machine would benefit from RAID (like a single rendering workstation, whether CAD, high-end video, or professional music studio). For the gamer or home user and even the typical office applications, RAID offers no speed benefits.

Now, when you can comment on what has actually been stated instead of arguing against what your incorrect assessment from not reading has stated, go ahead and get back to me.

[shieldforyoureyes]

RAID was designed to beat the speed & reliability of mainframe
drives, on mainframes, at a fraction of the cost. The Patterson
article criticizes mainframe & minicomputer drives for having
fallen behind in capacity and speed. They were not the goal
to reach, they were the problem to solve.
If you say "RAID doesn't give you speed" to refer to anything
other than the io speed of the RAID device, you're begging
to be misunderstood.
I'll accept that the comment about "redundency" and multi-user
environments was a very loosely worded description of IO issues
which I read too literally. Sorry.

[GreNME]

Well, glad that's cleared up. :)

[shieldforyoureyes]

The thing I don't get is the anti-RAID0 crowd never seems
to be against buying new fast drives. Doing RAID0 across
a couple of older 50 meg/sec drives "won't get you anything",
but replacing them with a newer 100 meg/sec drive is a
good idea?
Obviously there are other factors with new drives (price/
capacity), but would you claim that there is no performance
gain to be had for most PC users by replacing a 50 meg/sec
drive with a 100 meg/sec drive? (Or whatever speed the
latest & greatest is this week.)
(I would, I'm just asking because I never hear it.)

[GreNME]

Oh, I absolutely think getting newer and faster drives will increase performance, whether running a RAID setup or a single-disk setup. Also, for those interested in using RAID, the better the controller chip (or card, which has the chip on it), the better performance you will see.

[v3rt1g0]

I currently am, after reading some of the colorful posts, I broke my raid0 array and now have both 120's as separate drives.. =/

That's the last reason you shouldn't use raid. Don't let these bashers change your mind just because they're more rabid in their opinions.

Facts:
- RAID-0 is faster than a single disk, period.
The anandtech article exclusively used benchmarks that were hand picked to show no difference between a higher i/o array and a single drive. That much is obvious. It was aimed at "average home users". Any of those on these boards? Exactly. Certain people are just trying to make themselves feel better by trying to reinforce the rapidly-growing myth that a single disk is "just as fast". Don't believe a word of it.
- You will loose the data on both drives if one dies
- Your drive may "wear out" a little bit sooner in an array. Raptors have 5 year warranties, so who cares.

[shieldforyoureyes]

Oh, I absolutely think getting newer and faster drives will increase performance, whether running a RAID setup or a single-disk setup. Also, for those interested in using RAID, the better the controller chip (or card, which has the chip on it), the better performance you will see.

Disk performance, or user-experience performance? Would you
tell a gamer that going from a 50 meg/sec drive to a 100
meg/sec drive will affect their gaming experience in any
way?

[GreNME]

Disk performance, or user-experience performance? Would you
tell a gamer that going from a 50 meg/sec drive to a 100
meg/sec drive will affect their gaming experience in any
way?
User-experience performance will not be significant, though disk performance will be. The average gamer will generally see somewhere between a 2-5% increase, which may get them a frame or two per second more and a higher score on synthetic benchmarks, but not a noticably snappier experience. The improvement will be there, and can be measured, but not through playing Doom3.

The anandtech article exclusively used benchmarks that were hand picked to show no difference between a higher i/o array and a single drive. That much is obvious. It was aimed at "average home users". Any of those on these boards? Exactly. Certain people are just trying to make themselves feel better by trying to reinforce the rapidly-growing myth that a single disk is "just as fast". Don't believe a word of it.
Ha! Conspiracy theories abound!

[shieldforyoureyes]

There we go - my whole problem with this debate is that people
who think that increased disk performance will not help the
average desktop user are saying "RAID isn't fast!", which I
read as a statement about RAID, not about storage IO in
general.
Has anyone made a disk that does RAID5 across the platters
of a single drive yet?
There was one company that did RAID across tape drives, which
is kind of insane.

[liquidtrance123]

That's the last reason you shouldn't use raid. Don't let these bashers change your mind just because they're more rabid in their opinions.

Facts:
- RAID-0 is faster than a single disk, period.
The anandtech article exclusively used benchmarks that were hand picked to show no difference between a higher i/o array and a single drive. That much is obvious. It was aimed at "average home users". Any of those on these boards? Exactly. Certain people are just trying to make themselves feel better by trying to reinforce the rapidly-growing myth that a single disk is "just as fast". Don't believe a word of it.
- You will loose the data on both drives if one dies
- Your drive may "wear out" a little bit sooner in an array. Raptors have 5 year warranties, so who cares.
QFMFT
thank you, i thought i was the only one on these boards that felt this way :) :D :o :cool:

[liquidtrance123]

User-experience performance will not be significant, though disk performance will be. The average gamer will generally see somewhere between a 2-5% increase, which may get them a frame or two per second more and a higher score on synthetic benchmarks, but not a noticably snappier experience. The improvement will be there, and can be measured, but not through playing Doom3.


Ha! Conspiracy theories abound!
I agree overall performance won't be "astounding" in most cases, my point was the actual disk performance was much better with my Raid0 array than with only 1 drive. When i put my disks in Raid i wasn't looking for a FPS gain or anything like that, more so for just file transfers.

Plus bragging rights :p :D

[shieldforyoureyes]

It's not so much as "average home user" as "average home use".
If you're mostly gaming, you're probably not limited by disk
io speed. You may have a liquid-cooled, overclocked, dual
processor rig, and you might still not be maxing out your
disk speed.
Maybe.
My first response to any question about RAID is, "What are
you doing?".
I have a 30 disk fibre channel storage array* on my main
computer, and I use RAID heavily. It's good for me, it's
good for lots of people, it isn't good for everyone.
[* - Currently half-populated.]

[GreNME]

It's not so much as "average home user" as "average home use".
If you're mostly gaming, you're probably not limited by disk
io speed. You may have a liquid-cooled, overclocked, dual
processor rig, and you might still not be maxing out your
disk speed.
Maybe.
My first response to any question about RAID is, "What are
you doing?".
I have a 30 disk fibre channel storage array* on my main
computer, and I use RAID heavily. It's good for me, it's
good for lots of people, it isn't good for everyone.
[* - Currently half-populated.]
^ That is a good answer. (bolded text for emphasis)

Lots of rendering or dealing with huge I/O requirements—RAID will likely be helpful
Playing Battlefield: 1945 or Doom3—RAID will likely not make a difference
Using Office, surfing web, watching DVDs—RAID will likely not make a difference

And so on, and so forth...

[TheGameguru]

That's the last reason you shouldn't use raid. Don't let these bashers change your mind just because they're more rabid in their opinions.

Facts:
- RAID-0 is faster than a single disk, period.
The anandtech article exclusively used benchmarks that were hand picked to show no difference between a higher i/o array and a single drive. That much is obvious. It was aimed at "average home users". Any of those on these boards? Exactly. Certain people are just trying to make themselves feel better by trying to reinforce the rapidly-growing myth that a single disk is "just as fast". Don't believe a word of it.
- You will loose the data on both drives if one dies
- Your drive may "wear out" a little bit sooner in an array. Raptors have 5 year warranties, so who cares.


:confused:

First you say that Raid 0 is faster than a single disk PERIOD.

Period means no exceptions.. then you quote where there are exceptions.

Which one is it? Is it faster or not?

Lets face it... if your a general user or a gamer one 74 gb Raptor is going to be slightly faster than 2 74gb Raptors in Raid 0.

The upsides outweigh the downsides.

i.e unless your doing very specific tasks where Raid 0 will show you tangible benefits (and there are some) its not worth the added risk/performance hit.

If you want to brag and boast because you think its cool then thats a whole other issue.

I'd like to think that here were a little beyond the average Maximum PC crowd..guess I'm wrong.

[GreNME]

I'd like to think that here were a little beyond the average Maximum PC crowd..guess I'm wrong.
I think that, in general, the [H] crowd is pretty damn decent. There are plenty of cats here that deserve mucho kudos for their contribution and efforts. On the other hand, just like anywhere, there are some who would rather take the easy road and regurgitate things they read on other sites without adequate research and testing (i.e.—"works for me!"). Don't let this lower your opinion of the [H] as a whole. :)

[phaelinx]

The only reason I broke my array was because of
1. I dont do heavy user load server stuff.
2. My 2 120gig hds are for storage and I only did it in the first place to have it seem cool, in which I didnt really see a performance increase so I broke the array.
3. If I did lose 1 hd..there goes all my porn..

I mainly wanna increase speed because I think I have a decent speed pc with nice ram and a nice cpu but I HATE sitting waiting for levels to load in games..thats the #1 thing I use this pc for..and having both those with a slow ass hd is my bottleneck in my system...

I've come to the conclusion that I want to get like a 15k rpm hd, but what one do you guys recommend?

[Snugglebear]

Seagate.

[Bling]

this is better than reading nerds post and fight over videocards

[Bling]

F-22 raptors are cool, so are the ones in Jurassic Park

[djnes]

First off, I'm not anti-RAID. It has it's advantages...IN THE RIGHT ENVIRONMENT. For example, I have a Proliant server in my basement running a RAID5 array. I wouldn't dream of NOT using that. Second, I have a file server at work that I use to share point for my group. I am currently running 2 WD 120 GB drives in RAID1.

The point was RAID0 is pretty much useless in a desktop environment. Most sites and magazines are finally agreeing with this. The problem is, a lot of you refuse to read the articles or the tests,

This is what's known as a paradigm shift. For a while, everyone ho was a "power-user", me included, thought RAID0 was the best thing available for desktop drives. Now, some very respected web and print organizations are doing some "real world" testing, and finding out it's not any faster, and in some cases is slower than a single drive of the same kind. That brings me to the second flaw with many fo the arguments here. There's a huge difference between synthetic benchmarks and real-world benchmarks. If you don't believe me, ask around in the video card forum...or ask Kyle to discuss Futuremark.

I used to be one that said "I know my Raptors in RAID0 are faster, because I can feel it. I can just tell". As a test, I broke up my array, and used just one of the Raptors. I did not notice any differences in speed. I used to clean Ghost images to quickly switch back and forth between RAID0 and a single drive. It's at that moment you realize how bullshit and ridiculous it was that you could say it "feels" faster. That's as un-scientific as it gets.

Now we can continue to go back and forth on this. But the fact remains. Some of you are arguing with fact. It's been proven, and beaten down. The fact that many of us long-time members are trying to encourage others to not use RAID is because we don't enjoy seeing a ton of threads with people asking how to recover data from a failed array. Me personally, I never stored and valuable data on my RAID0 anyway. If your going to have a wonderful backup solution, and your willing to buck the trend, install RAID0 and be happy. Just don't take the foolish route by telling people it's faster....it's not.

Sticking a big red R on your car doesn't make it faster, and neither does RAID0 in a computer.

[liquidtrance123]

Now we can continue to go back and forth on this. But the fact remains. Some of you are arguing with fact. It's been proven, and beaten down. The fact that many of us long-time members are trying to encourage others to not use RAID is because we don't enjoy seeing a ton of threads with people asking how to recover data from a failed array. Me personally, I never stored and valuable data on my RAID0 anyway. If your going to have a wonderful backup solution, and your willing to buck the trend, install RAID0 and be happy. Just don't take the foolish route by telling people it's faster....it's not.


:rolleyes:
sorry that i "thought" i noticed a difference i'll be sure to run my "thoughts" by you whenever i get another one, to make sure if its "right" or not.
:rolleyes:

edit: just to make sure we're on the same page here, your telling me i was dreaming when i noticed a difference correct? Kuz i'm really tired of arguing whether or not i'm lying when i said i noticed a difference.

[djnes]

:rolleyes:
sorry that i "thought" i noticed a difference i'll be sure to run my "thoughts" by you whenever i get another one, to make sure if its "right" or not.
:rolleyes:

edit: just to make sure we're on the same page here, your telling me i was dreaming when i noticed a difference correct? Kuz i'm really tired of arguing whether or not i'm lying when i said i noticed a difference.


Since your obviously reading comprehension impaired, I will state it for you once again. Your "noticed" difference was purely psychological. We've all done, thinking it made a difference. Now that we've run the actual tests, we have scientific data to prove it's not. When we switch back to a single drive, we don't notice any decrease.

Now, as much as myself and plenty of others are having a good time reading your ridiculous responses, how much longer are going to argue with fact. In the science community, we often call it the most telling sign of ignorance, when someone continues to argue, despite being proven wrong. Roll your eyes...do whatever childish things you may....all we're going to do is laugh more. Of all my years here, you are clearly the most childish I've come across. Sarcasm was pretty funny when we were in 5th grade. Grow up, research a little, be a man to admit you were wrong, and go about your day.

[liquidtrance123]

Since your obviously reading comprehension impaired, I will state it for you once again. Your "noticed" difference was purely psychological. We've all done, thinking it made a difference. Now that we've run the actual tests, we have scientific data to prove it's not. When we switch back to a single drive, we don't notice any decrease.

Now, as much as myself and plenty of others are having a good time reading your ridiculous responses, how much longer are going to argue with fact. In the science community, we often call it the most telling sign of ignorance, when someone continues to argue, despite being proven wrong. Roll your eyes...do whatever childish things you may....all we're going to do is laugh more. Of all my years here, you are clearly the most childish I've come across. Sarcasm was pretty funny when we were in 5th grade. Grow up, research a little, be a man to admit you were wrong, and go about your day.
say and believe what you will, but i DID notice a difference from 1 drive to 2. If you wanna sit here and argue till we're both blue in the face thats fine. But i'm not going to tell you i didn't Notice a difference. If you wanna say i'm childish for trying to say i noticed a difference thats fine. But i will not sit here and say your right and i didn't notice a difference in the raid. I'm now done with this thread and your going on my ignore list as i don't need to listen to people tell me what i did and didn't experience.

[djnes]

say and believe what you will, but i DID notice a difference from 1 drive to 2. If you wanna sit here and argue till we're both blue in the face thats fine. But i'm not going to tell you i didn't Notice a difference. If you wanna say i'm childish for trying to say i noticed a difference thats fine. But i will not sit here and say your right and i didn't notice a difference in the raid. I'm now done with this thread and your going on my ignore list as i don't need to listen to people tell me what i did and didn't experience.

I called you childish because you had to resort to childish sarcasm and little kiddie games, instead of discussing the topic with me. I'm glad your done with this thread...I was about to need a shovel to get through all the bull you were laying down.

[v3rt1g0]

The point was RAID0 is pretty much useless in a desktop environment. Most sites and magazines are finally agreeing with this. The problem is, a lot of you refuse to read the articles or the tests,

This is what's known as a paradigm shift. For a while, everyone ho was a "power-user", me included, thought RAID0 was the best thing available for desktop drives. Now, some very respected web and print organizations are doing some "real world" testing, and finding out it's not any faster, and in some cases is slower than a single drive of the same kind. That brings me to the second flaw with many fo the arguments here. There's a huge difference between synthetic benchmarks and real-world benchmarks. If you don't believe me, ask around in the video card forum...or ask Kyle to discuss Futuremark.

I used to be one that said "I know my Raptors in RAID0 are faster, because I can feel it. I can just tell". As a test, I broke up my array, and used just one of the Raptors. I did not notice any differences in speed. I used to clean Ghost images to quickly switch back and forth between RAID0 and a single drive. It's at that moment you realize how bullshit and ridiculous it was that you could say it "feels" faster. That's as un-scientific as it gets.

Now we can continue to go back and forth on this. But the fact remains. Some of you are arguing with fact. It's been proven, and beaten down. The fact that many of us long-time members are trying to encourage others to not use RAID is because we don't enjoy seeing a ton of threads with people asking how to recover data from a failed array. Me personally, I never stored and valuable data on my RAID0 anyway. If your going to have a wonderful backup solution, and your willing to buck the trend, install RAID0 and be happy. Just don't take the foolish route by telling people it's faster....it's not.

I don't refuse to read the articles, I refuse to agree with them.
You think just because they are published that they are absolute truth?

The Anandtech article used only TWO real world tests, both were games, and both *WERE TIMED BY HAND*. Yet you take the article as complete gospel as to why Raptors in RAID 0 are completely pointless? No Photoshop benchmarks? Video Editing? File copies? File Moves? System boot time? No HD tach throughput tests? Extracting files? Installing programs? Searching for files? Drive optimization?

Where's the beef?

You also say "There's a huge difference between synthetic benchmarks and real-world benchmarks. If you don't believe me, ask around in the video card forum...or ask Kyle to discuss Futuremark"
Yet you reference the Anandtech article which uses synthetic benchmarks almost exclusively.

None of us are arguing with fact, because what you post is not fact, it's opinion, synthetic benchmarks, and real world benchmarks that were timed by hand. You refer to articles that used a very narrow battery of tests which one cannot possibly draw a all encompassing conclusion from (but you have).

You make everyone using RAID 0 out to be forgotten hardcore user wannabes with their heads in the sand. Yet you adamantly profess that RAID 0 could be slower then a single drive. The same speed, yes, in some tests. Slower?... you actually believe that for a second?

[djnes]

You make everyone using RAID 0 out to be forgotten hardcore user wannabes with their heads in the sand. Yet you adamantly profess that RAID 0 could be slower then a single drive. heh. The same speed, yes, in some tests. Slower?... you actually believe that for a second?

Using a test you suggested, such as HDTach, I have tested the seek times of my Raptor RAID to have higher seek times than one single Raptor. Read more than just Anandtech's article. Read the ones at StorageReview. These are some of the most trusted and respected sites in the industry. I'm not ignorant to only believe what I read. But when I read something...and then repeat the tests and results myself, I believe it. That's called the Scientific Method. If it's repeatable, it's believable. Now...you want to keep arguing this too or just drop it?

[v3rt1g0]

Higher seek times of? 0.5ms? 1ms? You didn't mention the throughput, which generally more than makes up for a tiny bit higher seek time.

What motherboard are you using? Non PCI bottlenecked SATA?

Maybe you could point me to the StorageReview articles you mention. I looked through their articles but couldn't find anything about RAID with Raptors.

[djnes]

Maybe you could point me to the StorageReview articles you mention. I looked through their articles but couldn't find anything about RAID with Raptors.


It's RAID0 in general, not just with Raptors. Throughput only is a huge difference in the synthetic benchmarks. Various other people on here have posted results timing file transfers, load times of game levels (coming off the HDD and not CD), and other tests to show there is no difference. My seek times varied anywhere from .3 ms to 2.5 ms in my tests. Others have been able to find the same. My point is...once again...from the top...the very minimal speed gains (if any, which is the debatable part) do not in any way warrant your reduction by half of the MTBF. When you break the array and go back to a single drive, you loose nothing in overall performance.

Here's the link. If you read through, I think you'll see they pretty much say it doesn't offer you much at all. The pros are nothing compared to the cons.

http://faq.storagereview.com/SingleDriveVsRaid0

[v3rt1g0]

...once again...from the top...

No need for the the belittling comment.

You mention a halving of the MTBF. Please provide proof of this.
I understand it halves the length of time your DATA will survive (as you have twice the chance of a SINGLE drive dying and taking ALL your data), but to half the life of EACH drive just because it's part of a RAID array? The different data access patterns wear it out? Doesn't make sense.

I agree you loose no _overall_ performance, but for high disk i/o operations like copying files, extracting files and the like, it's damn hard to believe that a single disk is just as fast. I'll test this for myself this weekend hopefully. Just need to make this week's backup before I break my array.

[v3rt1g0]

Throughput only is a huge difference in the synthetic benchmarks.

Why do you say that? I would think I'd be a huge difference for a medium-large file operations in windows explorer.

Also, you didn't answer:
What motherboard are you using? Non PCI bottlenecked SATA?

[djnes]

Why do you say that? I would think I'd be a huge difference for a medium-large file operations in windows explorer.

Also, you didn't answer:
What motherboard are you using? Non PCI bottlenecked SATA?

Yeah, I did forget that question. I tested on two separate boards..Abit IC7 and the Abit AI7. Both of which do not use PCI-bus SATA ports.

One of the games commonly tested was BF:1942. It was done rather well. A few people got the no-CD cracks, so everything had to come from the HDD. Same games, same patch levels...and they timed the level loads of specific battles. If you go through the game directories, some of these files can be rather medium to large in size.

Look, I'm not trying to be a dick. It wasn't too long ago that I'd argue with people because my Sisoft Sandra scores were through the roof in my Raptor Raid. It's only been a month since I broke my array for good and kept them single. I don't even have them in the same PCs anymore. My points I was trying to make to liquidtrance123 was that it seemed like a great idea, you felt like it was faster, but when you did some actual operational testing...you were disappointed with the results. I'm not screaming that it's useless...just that it isn't giving you anything wonderful. It makes me worry more about my data, despite backing it up as well.