I think I NEED raid

[djnes]

Well said. And the more people that report back with their experience the better.

When various people have posted their results, you flame them, or draw out a long debate, basically telling them they are wrong. So now who's being hypoctrical? You've done that to quite a few people. You're attiude is what kills these threads, and takes it away from being a useful debate.

Here you actually state that it's good for people to post their experiences...but you forgot to say it's only good if they agree with you. You never accept a differing opinion, nor do you even except that others have differing results for RAID0. That is the very definition of being hypocritical. Then you accuse me of being hypocritical, when my stance on the issue has been uniform.

If you wish to be true to your comments above, allow others to say they see no benefit. I, myself, gave up a RAID0 array and split my Raptors between two computers. That providied no loss in performance for me, and in fact increased my value out of the money spent. No one will argue that a Raptor is faster than *most* single drives. At the time, my 36 GB Raptors where the fastest thing out there. Even the 74 GB ones weren't released yet. If you'd like to come up with some actual testing, and work with others on here, that would be welcomed and productive. I'd love to participate, but I no longer have two identical drives. My only array I have is a RAID1 array in my server.

 

[tuskenraider]

When various people have posted their results, you flame them, or draw out a long debate, basically telling them they are wrong. So now who's being hypoctrical? You've done that to quite a few people. You're attiude is what kills these threads, and takes it away from being a useful debate.

Here you actually state that it's good for people to post their experiences...but you forgot to say it's only good if they agree with you. You never accept a differing opinion, nor do you even except that others have differing results for RAID0. That is the very definition of being hypocritical. Then you accuse me of being hypocritical, when my stance on the issue has been uniform.

Again SHOW US where I've done these thing, since you consistantly claim I do this. You lied and were hypocritical and I stated so. You don't have the right to make false accusations as far as I'm concerned. The SEARCH function is up top. My attitide kills threads? What was my first post in this thread and what was yours? I'll show you:

If your interested in trying, 2 drives in RAID0 will speed up your boot time and many of your heavier program launches. I typically see about a 15-20% reduction in times. All the others that say it doesn't reduce loading, what were your results when you compared a single drive to RAID0 boot times in your system?

Here's a guide to RAID0 setup.
__________________

It would be nice, but his idea of fun is to flame people who don't agree with him on a certain topic, until the thread gets locked. So far, the advice others have given has been correct.

It's kind of like when everyone thought the world was flat. At some point, proof was given that it was round. Some choice to agree, and some chose to live under their rocks in the past. It's the difference between learning, and closed-mindedness.

Just ignore him. If he wants to use RAID0 and pretend it's faster, so be it.

To the OP: If you aren't willing to buy extra drives, you won't be cnosidering RAID anyway. Not that it would solve your situation, but you can't run RAID on a single drive. Worrying about boot time is somewhat of a waste anyway. Cutting your boot time from 30 second to 25 isn't really doing anything miportant, is it? Chances are, if you do have a very long boot time, it's a software or network issue anyway.

Who's post is unaccepting of anothers opinion? You want to keep this up go right ahead. Funny you're doing the very thing you claim I am and you're not.

If you wish to be true to your comments above, allow others to say they see no benefit. I, myself, gave up a RAID0 array and split my Raptors between two computers. That providied no loss in performance for me, and in fact increased my value out of the money spent. No one will argue that a Raptor is faster than *most* single drives. At the time, my 36 GB Raptors where the fastest thing out there. Even the 74 GB ones weren't released yet. If you'd like to come up with some actual testing, and work with others on here, that would be welcomed and productive. I'd love to participate, but I no longer have two identical drives. My only array I have is a RAID1 array in my server

I can't count the number of times I've busted up my array and try single drives everytime these types of threads come up. I actually let some of you almost doubt myself. I say to myself, "Am I 100% certain of my claimed benefits?" And the answer is always YES. I've tried the OS on one drive, apps on the other and various pagefile setups. Then I've tried OS and apps on one and just a pagefile on the other, etc. I time bootup, a few apps and file compressions and end up putting my array back together within a 1/2 hour.

 

[tuskenraider]

In your proposed solution? You're either delusional or lying. I read through all of your posts and the only place you mentioned Matrix was just now. You even posted this link
http://www.overclock.net/faqs/48758-...ers-guide.html
to help the guy set it up. No mention of Matrix there is there???

Apparantly you missed post #28. My first post you partly quoted was just to say "Hey, I've had this kind of experience with RAID0, here's a link." I did that hoping to avoid the very thing we are doing now.

You did not talk about any of the drawbacks of RAID-0 or his potential for data loss, higher CPU utilization, higher latency than a Raptor, etc. That is completely irresponsible. For all you know, he could have his daughters first steps video going on his machine, important school work or things he actually makes a living from. I realize that you're a RAID-0 zealot. In some cases, the cost can actually make sense when you don't need high availability and you do have a backup scheme. To advocate it to someone who doesn't know about it without discussing the other points is not something you should be doing.

Everyone else covers the drawbacks just fine. As far as CPU utilization, what modern CPU even flinches at what an onboard controller uses? What program are you using to claim high CPU utilization? If it's HD Tach, why does it say 8%, but when I watch Task Manager and access the array, I see no such usage? Unless you're on a 500mhz Celeron, it's a non-issue. As far as latency, he said he would buy 2 more drives, which could be Raptors. I used adding another 320GB in Matrix as an example of the cheapest setup he could put together. You call me a zealot based on what? Responding to the naysayers? I've never stated everyone should have RAID0. If you don't think my posts provide enough information to your liking, you can certainly fill in what you deem neccessary.

 

[tuskenraider]

There are 2 kinds of RAID 0 users:

- Those with a functioning RAID 0 array that will eventually fail,

- Those that lost their RAID 0 array already.

Anyone with a hard drive falls under the same umbrella unfortunately. I find it interesting that I've lost 2 data drives in the last 5 years but all of my RAID0 setups have been problem free.

 

[general]

Great, now you're insinuating that RAID-0 leads to an increase in hard drive reliability. You can say that you left it up to other people to explain why this guy should have made RAID-0, but in your first post to him you did not and have had not. You should never tell anyone to do that without explaining the risks, tradeoffs and benefits. There are drawbacks even with a Matrix RAID setup. Yes CPU utilization is higher (even if only a few percent) and yes you do lose availability.

You come on here acting like RAID is a panacea and it's simply not. It's great that you enjoy it, but if you're going to advocate it, please be responsible with what you do. I don't see you hopping into threads where people have lost all of their data telling them how they'll magically get it back. I do see that you commonly advocate people running RAID-0 though.

 

[Dan_D]

Anyone with a hard drive falls under the same umbrella unfortunately. I find it interesting that I've lost 2 data drives in the last 5 years but all of my RAID0 setups have been problem free.

There are too many variables involved in hard drive failure for us to know the reason why your drives failed. The simple fact is that RAID-0 is NOT a reliable means of storing data. You have two failure points and therefore you are twice as likely to lose data than you would be using one drive. I do not see how anyone could possibly conclude differently. The only reason to run RAID-0 is because you simply don't care about data loss and you want the performance improvement you think you will get.

The fact is that software/onboard RAID-0 isn't any faster in real world application than a single drive setup is. (Outside of synthetic tests) In my experience the only "improvement RAID-0" has ever given me is reduced level load times in rare instances in some games. (Far Cry seems to really like RAID-0.)

BTW guys RAID-0 actually can increase boot times due to the RAID controller BIOS initialization time for the RAID. So the total time it takes to get to the Windows desktop from the moment you power the system on may actually be longer than the time taken by single disk systems with no RAID controllers/functionality enabled to do the same. So any boot speed advantages are likely negated by extra time needed for the RAID BIOS to initialize.

I'm running a RAID-0 setup now, but in the future I might not be. I've been considering getting a single Western Digital Raptor 150 drive and ditching my two 74GB units. I've also considered keeping the RAID array, but moving it to a real hardware based controller card that is PCI-Express based. IMHO that is the only way to do RAID right.

 

[tuskenraider]

Great, now you're insinuating that RAID-0 leads to an increase in hard drive reliability. You can say that you left it up to other people to explain why this guy should have made RAID-0, but in your first post to him you did not and have had not. You should never tell anyone to do that without explaining the risks, tradeoffs and benefits. There are drawbacks even with a Matrix RAID setup. Yes CPU utilization is higher (even if only a few percent) and yes you do lose availability. If you don't want to respect my opinion for that, I'll be just fine.

You come on here acting like RAID is a panacea and it's simply not. It's great that you enjoy it, but if you're going to advocate it, please be responsible with what you do. I don't see you hopping into threads where people have lost all of their data telling them how they'll magically get it back. I do see that you commonly advocate people running RAID-0 though.

I've insinuated nothing. I just noted how ironic that with the increased risk of RAID0, it's the single drives that let me down. I think I could find that 80% of the posts that lack one of the three catagories you mentioned. Noone is required to post to your expectations. Everyone provides their little peice of the puzzle in their own way. I think it's something you cannot change and will have to accept. The CPU utilization is a non-issue, but if you can post a link proving otherwise, I'd love to read it. I could shut down unused Windows processes and get that back. As far as helping people with data loss, it's my choice to respond to a thread or not, and I don't need create a "balance" in my thread response selection. If that means you don't respect my opinion, I'll have to live with it.

There are too many variables involved in hard drive failure for us to know the reason why your drives failed. The simple fact is that RAID-0 is NOT a reliable means of storing data. You have two failure points and therefore you are twice as likely to lose data than you would be using one drive. I do not see how anyone could possibly conclude differently. The only reason to run RAID-0 is because you simply don't care about data loss and you want the performance improvement you think you will get.

The fact is that software/onboard RAID-0 isn't any faster in real world application than a single drive setup is. (Outside of synthetic tests) In my experience the only "improvement RAID-0" has ever given me is reduced level load times in rare instances in some games. (Far Cry seems to really like RAID-0.)

BTW guys RAID-0 actually can increase boot times due to the RAID controller BIOS initialization time for the RAID. So the total time it takes to get to the Windows desktop from the moment you power the system on may actually be longer than the time taken by single disk systems with no RAID controllers/functionality enabled to do the same. So any boot speed advantages are likely negated by extra time needed for the RAID BIOS to initialize.

Noone here has called RAID0 reliable, but at the same, it's not unreliable either. That makes it a non-issue for me. Citing lost data posts to me does nothing but show people with problems seek help, and those without, don't. I won't argue about my proven performance increases anymore as they've been stated. As far as increased boot times with a RAID based controller, even if you're are not in a RAID configuration but use the ports controlled by it in IDE mode, the initialization time is the same. So if you really want to save time, buy a motherboard that doesn't have its drive ports connected to a RAID based controller. Unfortunately, I think most of the better enthusiast motherboards(overclockability, etc.) all have RAID controllers.

 

[FlatLine84]

On the reliability note.... I'm of the camp to believe that it's about the same in reliability, at least for home/enthusiest user. I do understand that RAID 0 increases the chance of data loss due to hard drive faliure by more than 2x. But anytime you buy a hard drive, you are rolling the dice on it's reliabilty. Today though with reputible brands, it seems to be very rare to get a bad drive. Obviously if you look across a broad range of users, the failure rate will be higher, and say in like a buisness, that wouldn't be so good. But on an individual perspective, it will be much less. Meaning for myself, my backup drive has failed before my RAID 0 array I've been running for 4 years. I may be able to have 10+ installs before a failure. Others it may be different. There is no question that the amount of "luck" goes down with 2 drives in RAID 0, but it's still a gamble. No one can say you're screwed by running a RAID 0 array.

 

[drizzt81]

I'm of the camp to believe that it's about the same in reliability, at least for home/enthusiest user. I do understand that RAID 0 increases the chance of data loss due to hard drive faliure by more than 2x. But anytime you buy a hard drive, you are rolling the dice on it's reliabilty.

So you are saying because HDD failure is a random process, we do not need to worry about probabilities? That is an interesting thesis. In the end, you are not too far off: Given that "home users" generally have a small number of samples, their estimates for HDD reliability are unlikely to be representative for the population.
To put if differently: The behavior that home users observe, let's call it $h^{\prime}_{i}$, may appear to belong to a different probability distribution than the one that actually governs HDD failure rates, $h$. Given that $h^{\prime}_{i}$ is randomly drawn from $h$, we can say that $\sum_{\forall i}{h^{\prime}_{i}} = h$. This implies that because you do not see any failures, there has to be someone else, who sees ``excessive'' failures, so that $E[h^{\prime}] = E[h]$. Now we have come full circle, the expected observed failure rate is equal the to expected real failure rate. As such, it does not matter whether you are a home user or a business user, it matters how expensive downtime, possibly data loss, is.

Pretty version
Latex source

 

[FlatLine84]

So you are saying because HDD failure is a random process, we do not need to worry about probabilities? That is an interesting thesis. In the end, you are not too far off: Given that "home users" generally have a small number of samples, their estimates for HDD reliability are unlikely to be representative for the population.
To put if differently: The behavior that home users observe, let's call it $h^{\prime}_{i}$, may appear to belong to a different probability distribution than the one that actually governs HDD failure rates, $h$. Given that $h^{\prime}_{i}$ is randomly drawn from $h$, we can say that $\sum_{\forall i}{h^{\prime}_{i}} = h$. This implies that because you do not see any failures, there has to be someone else, who sees ``excessive'' failures, so that $E[h^{\prime}] = E[h]$. Now we have come full circle, the expected observed failure rate is equal the to expected real failure rate. As such, it does not matter whether you are a home user or a business user, it matters how expensive downtime, possibly data loss, is.

Pretty version
Latex source

Hehehe, exactly. I'm not discounting the fact the failure rate is there, and you do in fact mroe than double the chances of data failure with RAID 0, but more that I don't care, I'm rolling the dice either when I turn on my computer, and I use good backup practices. I just think that if someone wants to use it, and they know the potential problem, let em do it.

 

[Dan_D]

I've insinuated nothing. I just noted how ironic that with the increased risk of RAID0, it's the single drives that let me down. I think I could find that 80% of the posts that lack one of the three catagories you mentioned. Noone is required to post to your expectations. Everyone provides their little peice of the puzzle in their own way. I think it's something you cannot change and will have to accept. The CPU utilization is a non-issue, but if you can post a link proving otherwise, I'd love to read it. I could shut down unused Windows processes and get that back. As far as helping people with data loss, it's my choice to respond to a thread or not, and I don't need create a "balance" in my thread response selection. If that means you don't respect my opinion, I'll have to live with it.
Noone here has called RAID0 reliable, but at the same, it's not unreliable either. That makes it a non-issue for me. Citing lost data posts to me does nothing but show people with problems seek help, and those without, don't. I won't argue about my proven performance increases anymore as they've been stated. As far as increased boot times with a RAID based controller, even if you're are not in a RAID configuration but use the ports controlled by it in IDE mode, the initialization time is the same. So if you really want to save time, buy a motherboard that doesn't have its drive ports connected to a RAID based controller. Unfortunately, I think most of the better enthusiast motherboards(overclockability, etc.) all have RAID controllers.

Noone isn't a word. The proper usage is "no one".

You are also incorrect in that the controllers only go through RAID BIOS initialization if the onboard drive controller is set to RAID mode in the BIOS. If it is not, then only one drive initialization sequence occurs. Having RAID disks and enabling extra controllers does increase total boot time of the system.

 

[tuskenraider]

Noone isn't a word. The proper usage is "no one".

You are also incorrect in that the controllers only go through RAID BIOS initialization if the onboard drive controller is set to RAID mode in the BIOS. If it is not, then only one drive initialization sequence occurs. Having RAID disks and enabling extra controllers does increase total boot time of the system.

Thanks for the word tip, typos happen. Well with the last 4 boards I've had, I still had to wait for the initialization of the controller to appear on the boot screen before boot proceeds in IDE mode. My current board doesn't, so I'll give 4 out of 5 the majority. If boot is sped up by 1/2 second in IDE mode, I hadn't noticed it with those 4 boards.