Do I need to format to RAID-0

[madtownidiot]

I don't put much stock in benchmarks either, although the benchmarks for a pair of WD RE4s in RAID 0 consistently outscore most SSDs, but then most people aren't willing to spend $389 each for a pair of HDDs. Only an idiot would try to convince me, however, that my system is slower in any way than it would be if the two HDDs weren't set up in RAID. Of course if one fails, I lose the data on both, but that's why I paid extra for enterprise HDDs that are rated for 1.2 million hours and have a 5 year warranty. In real world applications, the HDDs outperform a single SSD in nearly every perceptible way. For example, it only took 40 minutes to install windows, vs more than an hour to install to an SSD (which I did try 1st while waiting for the HDDs to arrive). Games and VMs install and load faster, files can be moved from one partition to another faster than from an SSD to the HDDs. It took less than 3 hours for WMP to add 290 GB of music and 1300GB of videos to the library, something not even practical with a SSD. Obviously, someone isn't going to get the same results from a pair of HDDs off the shelf at best buy. But if you're willing to pay more, you can get comparable performance to an ssd and a hell of a lot more storage space for the money.

 

[DeaconFrost]

Only an idiot would try to convince me, however, that my system is slower in any way than it would be if the two HDDs weren't set up in RAID.

First time around you used moron, now I guess I am just an idiot? Higher seek times will lower performance in some aspects. If you don't want to believe me, here's the link to the article that ended the RAID0 debate once and for all. It's several years old, but that just shows how long it's been dead.
Western Digital's Raptors in RAID-0: Are two drives better than one? - AnandTech :: Your Source for Hardware Analysis and News

But if you're willing to pay more, you can get comparable performance to an ssd and a hell of a lot more storage space for the money.

Is this a good time to tell you that those drives cost more because they are meant to last longer running 24x7...and has nothing to do with performance? I also would question your previous comments about their reliability, because any drive can fail, regardless of the marketing behind it. I bought a brand new Dell PowerEdge R710 to run VMWare's ESXI server for my company, replacing 4 physical servers, and 2 of the 6 drives have died already, in the first three months of operation. Those are expensive, enterprise-class drives. There's no reason to shell out that extra cash for home use...you aren't guaranteeing anything. The enterprise drives have a longer warranty typically, but they aren't guaranteed against failure...that would be impossible.

For example, it only took 40 minutes to install windows, vs more than an hour to install to an SSD (which I did try 1st while waiting for the HDDs to arrive).

And now we may have gotten to the root of your problem. You either have an issue with the SSD, or you picked a very poor performing one. If it took you an hour to install Windows 7 on an SSD....something was wrong. How can I, and most others, install Windows 7 to a single mechanical drive in 15 minutes or less? My guess is, you had one of the above two issues, and decided to argue with me based on your one example...a sample size of 1. You do know that there are some early SSDs out there that had horrible performance, right? If you had no issues, and were using one of the better performing SSDs, you would never be telling me that a mechanical drive array would be faster. I recently set up a laptop for a friend of mine using an OCZ Vertex SSD. I had Windows 7 Pro installed from a blank drive in 8 minutes.

My suggestion? Read the article linked above to get the facts. I'm glad you are happy with your enterprise drives, and I do hope you are smart enough to have all that data backed up, rather than trusting them to the array only. However, if you do read the article, and then maybe some follow up topics, I think you'll be surprised with what you find out.

 

[chev65]

The misinformation contained in this thread is simply unbelievable. Since when does a single bench mark define the performance of any piece of hardware? And if you don't think SSD's are faster than standard drives you are sadly mistaken. Please spare me the reply guys, I've seen and read through just about every bench mark for HDD's there is. They are boring as heck and the difference between drive arrangements isn't that big anyway.

 

[DeaconFrost]

And if you don't think SSD's are faster than standard drives you are sadly mistaken.

I think the problem is, he had a bad SSD or one of the early, piss-poor performance ones. I don't like to get in the habit of defending people who refer to me as an idiot or moron...but I think in this case, he at least was basing his opinions off of an issue.

I do agree with you, and that's why these threads always make me scratch my head...because RAID threads are always filled with misinformation. Some of the misinformation is moldy and rotten, due to the length of time since it's expiration.

 

[madtownidiot]

Seems kind of foolish to form an all encompassing opinion on RAID based on an article that tested a single (now obsolete) RAID setup more than 6 years ago, to me at least.

And yeah.. seek times are a bit slower, according to the benchmark, but can anyone even tell the difference, honestly? And what about the assumption that somehow using two HDDs in raid doubles your chance of a system failure? May look that way mathematically, but numbers and statistics don't always apply to the real world. All parts eventually fail. That's a given. Sometimes you get a few bad parts, or even a whole lot of them, and if you're working with Dell, chances are they're going to replace the bad parts with more of the same.. they've done it before. Show me an article that has tested several RAID setups with a variety of hardware and gets the same results and maybe I'll start paying attention, but I've been using RAID in gaming systems for years and only once had a HDD fail in a way that I couldn't recover. None of them were any faster with a single HDD setup in any way.

I bought enterprise class HDDs for my build BECAUSE I needed a system that I can keep running 24/7 for at least the next couple years. You can't tell me point blank that RAID 0 makes a computer slower in any way if it's done right.

The SSD I was using is a 256GB OCZ Vertex by the way.. it worked great as a pair in the laptop I was using until about two weeks ago for my work, but by itself has a slower sustained read/write speed than my two mechanical drives. And installing windows in this case meant installing windows, drivers, and updates.

I didn't intend my last post as a personal attack, but really this whole thread has turned a bit moronic

 

[DeaconFrost]

And what about the assumption that somehow using two HDDs in raid doubles your chance of system failure?

I knew this was coming. Each and everytime, basic math goes right out the window. You're entire array, in RAID0 depends on each and every drive functioning. There is zero redundancy. Therefore, the more drives you have, the greater chance of failure. Two drives double the failure risk over one single drive. Three drives in the array? Triples the risk. It is a very basic principle of probability and statistics. I'm not going to waste any more time debating this point...because we might as well debate the fact that 2+2=4.

you can't tell me point blank that RAID 0 makes a computer slower in any way if it's done right.

Read the article, and then read about seek times. It slows down certain operations, as I said above. Just read the article.

Yes, it is a six year old article, but that actually adds to my side of the debate. That was done pre-SSD. With the advent of SSDs, that only furthers my point. I'm not going to comment on arrays made of SSDs, as I haven't done any reading on the subject, and I don't like to make ASSumptions.

 

[madtownidiot]

It's not basic math.. Adding HDDS in RAID 0 doesn't mean your computer is going to fail any sooner, although the author of that article wants you to believe otherwise... All pointless because how the hell do you figure the chances of a single HDD failing anyway, let alone quantify the chances of one in 2 or 3 or even 20 HDDs failing? Of course there's no redundancy in a RAID 0 setup, but I'm not foolish enough to trust a single computer to hold the only copy of any file that is important to me, and people who don't back things up can hardly blame anyone but themselves if they lose data.

I have read the article. It was written about a single pair of hdds and a single raid controller, and applies to that particular set of hardware. In any case is not even close to indicative of all possible hardware combinations/ raid setups. The seek time test itself is not a practical test, because it's based on finding random, non sequential sectors as quickly as possible, which is hardly the way a HDD with a healthy file system on a well maintained computer works in everyday use, and if your HDDs are decrapified and defragmented, actual seek times for the files it takes to load windows and other applications is going to be much faster than any benchmark result will show on the same hardware. So basically, I'm saying without qualification that the article in question is bullsh*t

 

[ChiefRA]

Hi again guys,

First of all, DeaconFrost did you noticed that the linked article you posted here from AnandTech is from early 2004? I mean, come on, I don't want to argue with you, but we are 6 years later... the HDD's are WAY up in performance.

I will tell you a fact, simply what I did: I bought 2 x WD 640Gb Caviar Black EX SATA 3-64MB buffer (WD6402AAEX) because I didn't wan't to throw money out of the window on Raptors. They're not that much better than the latest Blacks, and they cost 4 times more (the Raptors): as we speak - in 20.08.2010 - talking about same sizes and connectors:
a) Caviar Black EX 640Gb SATA3-64MB - WD6402AAEX - costs 66 EURO/each at most
b) VelociRaptor 600Gb SATA3-32MB - WD6000HLHX - costs 308 EURO/each

So I've placed those 2 blacks into RAID0 and they literaly FLY on all applications. Besides, the most use I've got from them, is when I open 3-4 programs in background - and they run, I mean "eating" information from those HDD's - and I have never encountered the stuck problem I had before when trying to open another program and the single HDD was screaming like hell trying to feed the information for my new opened program while they are working with the other allready started programs. So yes, the performance is WAY better than one HDD only or non-RAID configuration. To jump to a PERSONAL conclusion:

HDD in RAID0 vs. non-RAID configuration:
1) RAID0 config is a lot faster than non-RAID when you do real multitasking applications - running multiple programs in the same time and/or editing/copying large files - simply because it splits the feeding information in 2.
2) NEVER rely only on RAID0 with only 2 HDD's - for your own data protection. If you have money, buy 4 HDD's and do a RAID10 (or RAID 0+1) and you will thank me for this.

A simple piece of advice: 4 HDD's WD6402AAEX of 640Gb - That's a RAID10 - costs 264 EURO's, that less than one sigle Raptor drive.
SDD Drives are out of discussion for now, they're just TOO EXPENSIVE right now.

ChiefRA

 

[DeaconFrost]

First of all, DeaconFrost did you noticed that the linked article you posted here from AnandTech is from early 2004? I mean, come on, I don't want to argue with you, but we are 6 years later... the HDD's are WAY up in performance.

Yes, they have increased in performance...but that increase is a constant for both sides of the debate, and therefore is irrelevant. That's a basic principle of the Scientific Method that we all learned in grade school and high school for handling experiments. The reason I pointed out that it was a six year old article...was to show that's how long it's been since this debate has been raging on....until this thread, apparently. Most other forums I participate on....it isn't really discussed anymore. Mechanical drive arrays are left for servers, while the debate is on SSD arrays. Also...as another final point...notice how the high end system manufacturers aren't using arrays...haven't been for a while...they are all about SSDs.

 

[GRAHAM3D]

Striped or Normal Formatting for RAID 0

OK. I have created an SSD RAID 0 in my ASUS BIOS.
DO I have to do the RAID again in Windows 7 "Disk Management"?
Has it already been Striped by the BIOS or do I have to format it as Striped?

 

[ignatzatsonic]

About all I know about RAID 0 is that it's pretty much pointless with SSDs.

But maybe you have some very unusual situation or inside information.