Should I Go RAID-0?

[jph1589]

I feel a real debate coming on here. I guess we'd better open another thread in the right place.

[One4yu2c]

Originally Posted by jph1589

I feel a real debate coming on here. I guess we'd better open another thread in the right place.

I'll do ya one better and split this topic so the discussion can pick up where it left off.

As the thread title suggests, RAID-0 or not to RAID-0? Sound off...

[jph1589]

From the anandtech article

"The main downside to RAID-0, other than cost, is reliability. The size of a RAID-0 array is the sum of all of its members; so, two 100GB drives in a RAID-0 array will give you one array with a 200GB total capacity. Unfortunately, if you lose any one of the drives in the array, all of your data is lost and isn't recoverable. Since two drives are working in tandem and are both necessary to hold your data, you effectively halve the mean time between failure by moving to a two-drive RAID-0 array."

I completely disagree with all of that. If you had a 200gb drive instead and it failed, you still lose all your data. Since their first objection was the cost of 2 drives, they must have been suggesting that you use one 200gb drive. So the mean time between failure does not change at all. MYTH BUSTED!! LOL

Another quote from Anand.... "When Intel introduced ICH5, and now with ICH6, they effectively brought RAID to the mainstream, pushing many users finally to bite the bullet and buy two hard drives for "added performance". While we applaud Intel for bringing the technology to the mainstream, we'd caution users out there to think twice before buying two expensive Raptors or any other drive for performance reasons. Your system will most likely run just as fast with only one drive, but if you have the spare cash, a bit more reliability and peace of mind may be worth setting up a RAID-1 array.

Maybe the problem with his tests was that he used an Intel ICH5R controller for it and an intel motherboard. I really don't know that, but I still have a bad taste in my mouth over that intel board i had in a previous life. I also think that the comment about using Raid 1 on a desktop is probably as ridiculous as he thinks that using Raid 0 is. His first arguement against raid 0 was cost, so why wasnt that objection there for raid 1? Raid 1 belongs on a server not a desktop. Raid 0 does not belong on a server, it belongs on a desktop.

With the way that windows and users segregate their data, they back up their important files if they really care about them. The people who really don't care about them don't. Those people who don't care about their data, wont be using raid 1. The people who do care, will not stand for a decrease in performance when they can copy My Documents to a CD-Rw every week instead of slowing their machine down with raid 1.

So when I know that running Raid 0 showed a 60 percent increase in performance on HD tach on my brothers machine, and he can see it and feel it, I guess I have to question the benchmarks and Anand's conclusions.

You know the bottom line on this is probably not unlike us overclocking our computers and getting to the point where we can't see the increase in performance unless we run a benchmark. And even then maybe an increase of 500 mhz might only make a 1 second difference on superpi. But we still do it.

If I can run Raid 0 and get a performance increase of even 20 percent I think I'll probably be doing it.

IN MY OPINION OF COURSE :)

[Tyreal]

We can discuss this all we want and we still won't be a clear conclusion. There are some people that swear by their RAID setups and others who claim that it has no place in desktop rigs.

But there is a solution. RAID 0+1 or 1+0.

[One4yu2c]

Originally Posted by jph1589

I completely disagree with all of that. If you had a 200gb drive instead and it failed, you still lose all your data. Since their first objection was the cost of 2 drives, they must have been suggesting that you use one 200gb drive. So the mean time between failure does not change at all. MYTH BUSTED!! LOL

The wording is funky, but the point remains. Running two hard drives increases your odds of a failure. It's not that RAID-0 is any less reliable in and of itself, but the fact that you have additional hardware. The more drives you buy, the greater your chance of picking one up with a hardware defect, or developing a hardware defect over time. Same thing with RAM and 4 sticks versus 2. The more sticks you buy, the greater your chance of picking up a bad one. Or buying cars - the more you buy and run, the greater chance you have of needing a mechanic.

That being said, it's up to the end user to decide how risky that really is in the real-world. And regardless, everyone with valuable data should be running backups on a frequent basis.

Originally Posted by jph1589

So when I know that running Raid 0 showed a 60 percent increase in performance on HD tach on my brothers machine, and he can see it and feel it, I guess I have to question the benchmarks and Anand's conclusions.

Not just Anandtech (which, quite frankly, knows their stuff, and that particular article was written by the site owner), but Storage Review, Maximum PC, etc. Synthetic benchmarks are nice and all, as are testimonials, but in the end it depends on the user. Your brother may be running tasks where RAID-0 particularly shines, or it may simply be a placebo effect. Put him in front of two equally equipped systems except one is running 2 drives in RAID-0 and one is a single drive setup, and he may or may not be able to tell the difference.

Originally Posted by jph1589

If I can run Raid 0 and get a performance increase of even 20 percent I think I'll probably be doing it.

This is exactly what I caution against as an unrealistic expectation, at least in terms of real-world performance and non-synthetic benchmarking. If you're expecting a 20 percent increase in real performance, which would be quite significant, you very likely are setting yourself up for disappointment.

[jph1589]

I certainly don't want to seem aruguementative, and I don't want to p--- anyone off, but this kinda reminds me when I found out my ICH5 couldn't do NCQ and I had an NCQ drive. The Intel people spent days trying to convince me that NCQ had no place on a desktop, and that I was much better off without it. Even though I understood completely how it worked they finally convinced me that I didn't want it. I think they just wore me down actually LOL.

I'll be trying that raid 0 when i get the chance, and I'll report back on my results.....imagined or otherwise.

I bought the Hitachi's that do NCQ by the way.......so by gum....raid 0 and ncq had better do more than warm my computer case LOL.

I am glad I am at the point where I can find all these things more humorous than I used to. Just keep in mind my motto.............He who dies with the most toys....wins.

[One4yu2c]

Originally Posted by jph1589

I certainly don't want to seem aruguementative, and I don't want to p--- anyone off

No worries dude, this is what a forum is all about - the exchange of ideas, opinions, and educated (hopefully) discussions. :)

[polobunny]

Originally Posted by One4yu2c

The wording is funky, but the point remains. Running two hard drives increases your odds of a failure. It's not that RAID-0 is any less reliable in and of itself, but the fact that you have additional hardware. The more drives you buy, the greater your chance of picking one up with a hardware defect, or developing a hardware defect over time. Same thing with RAM and 4 sticks versus 2. The more sticks you buy, the greater your chance of picking up a bad one. Or buying cars - the more you buy and run, the greater chance you have of needing a mechanic.

Word brother. Tech knows; the more hardware the more points of failure. :)

Really though, nothing stops you from trying, but in my opinion it's just playing with your luck running a system 24/7 with a stripped raid. It looked real cool to me when everyone did it, less cool when the placebo faded away when testing it with something else than HDtach. ;) This program tells you strictly HD performance, not the performance you'll gain doing your things.
Try something like Winstone 2004 or Futuremark Sysmark 2004 and you will quickly see that those roaring HDD's don't add very much to your overall system performance, in the order of 2 to 3% MAX. You will notice the biggest performance boost when compressing/uncompressing archives or doing such activies, something you might not do very often. Game wise, on a 15 second level loading it will cut 2 seconds and no bigger FPS. In Windows, virtually no performance boost for normal applications.
It's definitely something cool to do, a kid dream, but not something you need.

[Captain Crunch]

I was thinking about doing raid at one point but the (Real World Performance) isn't even noticeable..........
It's kinda like running two 7950s to play to play a 2 year old game ......sure the bench mite look good but your eye would never see the difference in a million years.

[Capper]

I've run two Raptor 150 in RAID), a single Raptor 150, and a Seagate 7200.10 750GB HDD.....you'll see a little bit of improvement in load times, but it just isn't worth it in typical desktop usage.
You're better off sticking with a 300-500GB SATA II HDD (16mb cache, 3.0GB/s), maybe another drive for storing movies, music, etc. Look for drives that support the new features that will make a difference, like NCQ, etc.