Erssa wrote:sanse wrote:raid 5 makes things a lot slower and having a good backup on a drive on another place is still necessary. all imho of course.
Raid-5 indeed makes things a whole lot faster, not slower. It has faster read performance then raid-0 array, however write benefits are only very little due to parity overhead. Basically, It has the benefits of both striping and mirroring. Having a backup is not necessary, if a drive fails raid-5 can still function, and it can rebuild the array when you replace the failed drive
Erssa wrote:Again I have to say, that I never claimed RAID 5 does any mirroring ... I wasn't misleading.
jasonb885 wrote:You were. You even admit as such.
Your claim that in that quote I claim that RAID 5 mirrors. I will now explain the logic, so maybe even you will understand.
Assumptions:
RAID 5 is A
RAID 1 is B
Mirroring is C
Benefits of mirroring = Fault tolerance = D
If B then C
If C then D
I claim:
"Basically, A has the benefits of C, because C => D, therefore A => D"
You claim that when I say "A => D":
"Because A => D, and because C =>D, and because B => C therefore A = B".
Cannot you see the fault in your logic? There is no causality or implication that A does C. Do I need to explain the basics of cause and effect?
I could draw a truth table about this, but it would be pointless since, it seems that you disagree with my opinion and therefore say I am wrong, not because there is fault in the logic I presented.
I think you had problems understanding what you read, so you've drawn wrong conclusions and made up implications that do not exist.
Or do you disagree with the logic I presented?
My point is people not familiar need not become confused. So I offered clarification.
It's a good thing you want to make sure people understand the difference between mirroring and parity.
But based on my comment you assumed and implied*(proof 1) that I didn't know what I was talking about.
Proof 1 wrote:No. RAID 5 is stripping with parity only. There is _no_ mirroring component to RAID 5. Perhaps you meant RAID 1+0 or RAID 0+1?
Even when I mention parity overhead in the same verse, which should imply that I know what parity is.
The funny part is, that after tha comment you say.
jasonb885 wrote:In either case, RAID 5 is slower than an identical number of drives in a RAID 0 configuration, as stripping on RAID 5 is n-1 for any given stripe, since the parity information is on the 'final' disk. For n disks, RAID 0 is simply n, not n-1 for reads.
This is pretty funny, because it proves you don't know how RAIDs stripe. I will tell you:
Source:
http://wiki.ittoolbox.com/index.php/Top ... ,3,4,5,0/1
"Every entire data set of 2 kbytes written (also, the pre-allocated space on the disk in which a set could be written) is called a "stripe". The number of stripes on an array set is basically: useable size of all the disks combined / stripe size."
You were simply wrong. RAID 0 with similar number of drives in RAID 5 are faster
only, if the size of the RAID 0 array is bigger then RAID 5. Similar number of similar drives in RAID 0 will result in more stripes, but only because the total size will be bigger. Your calculations for calculation number of stripes is total bs. Besides RAID 5 has no "final" disk, or even "final disk", parity is divided evenly on the disks. RAID 3 however has the final disk. It seems you were the one confusing people, not me.
Erssa wrote:Having many drives on raid-5 array is also benefitial because, the amount of parity data is counted (n-1/n) where n is the number of hardrives in array. So he uses only total of 300gb from the 2.4Tb for parity information.
Sadly your stripe calculation formula was pretty similar to the formula of parity data I presented, earlier, pointing out that I knew what RAID 5 and parity was.
Can you admit you were wrong?
What you were doing here was pulling my text out of context and bluntly implying that I didn't know what I was talking about, while giving false information of stripe calculations etc... I cannot believe, that you honestly could have misunderstood me so badly. So I draw the conclusion, that you tried to rectify me just so you could feel smug about yourself. I have to admit, I feel pretty smug now.
Where I work we have 1,000s of machines and and catastrophic hardware failures do happen.
Statistically speaking, it would be quite amazing, if you didn't have failures. However when I made this poll, my intentions were, that even home systems with only 1 computer could have a say in this poll. I can say, that in the 12 years I have owned computers, none of the four drives I have owned have failed on me.
Do you think you'd be winning the lottery or something? Hardly.
Isn't that the biggest reason for people playing lottery? I hope, that some day I will win, but of course I am aware of the chances of me winning are extremely small, 8/15380937 every week to be exact.
You see to be completely confusing the concept of redundancy and backup, so for the benefit of others I'll simply elaborate further.
Oh please. You just proved you haven't read what people say in this thread. Tibors brought these up, before you made your first reply.
Erssa wrote:Tibors wrote:It seems you already know one reason why your first statement is not true. RAID 5 also doesn't save you from:
* User errors - Oops I deleted the wrong file.
* Bugs in software - Ever seen a complex Word file (lots of OLE) where the second half of the document was automagically turned into gibberish? I have seen that too many times and was glad the company I worked for didn't think RAID 5 was enough.
* Virus infection
* Malicious users
* Your mobo dies and the model is out of production. Say bye to your data or good luck finding one second hand.
You made some pretty good points. But none of these is a big threat to me.
Tibors already brought these up. And I replied to him, pointing that I didn't find those risks big enough to concern me. So when I disagree with you on the number of safety needed, does that make me instantly wrong or unable to understand the difference between the two?
Closest I have said are:
Having a backup is not necessary, if a drive fails raid-5 can still function, and it can rebuild the array when you replace the failed drive.
or
Nope. If I have raid-5 there's no need for me to have external drives for extra backups. There's a point where the extra protection comes redundant. For me it comes when I have raid-1, raid-5 or an external backup, I don't need both. The possibility of 2 drives failing at the same time is so small, that I don't consider it a big risk. Even external back up won't save you, if your house burns down.
The first quote has an important word "necessary". Necessary is a synonym for "absolutely essential" or a "must have". I mearly pointed out, that it's not of the utmost necessity to have external backup, if you have RAID 5, certainly you can have it and must have it, if your data is so valuable, that the loss of it even under the most unlike circumstances is too big of a risk. This doesn't apply to me. Hence I disagree it is
necessary.
That said, some of those items are more applicable than others depending on if you're running true hardware RAID, true software RAID, for fake vendor BIOS driver based RAID. I'd consider the latter most susceptible to all items above.
I agree with your reasoning here. However I disagree with the last sentence. I respect your opinion, do the same for me, at least until you can prove me wrong, with some undeniable evidence. I have no trouble admitting, if I am wrong or in the wrong trails. I admitted to Tibors, that I wasn't aware that RAID 5 was mostly driver/software based. However admitting to that does not mean I don't know the princible of how RAID arrays work, it only meant, that I wasn't fully aware how onboard RAID 5 implements the RAID 5.
Erssa wrote:First of all, drop the FRAID. Let's use the official and less confusing on-board RAID 5 instead.
Because it isn't on-board RAID anything. It's software based RAID using a vendor driver with basic hardware hooks. It isn't true hardware RAID or software RAID. It's just junk. So, Fake RAID seems appropriate.
How do you expect anyone to believe you are impartial, with objective information, when you insist on using such a term. Just think what it would sound, if a defendant was instead called criminal during the trial. Calling it FRAID shows prejudice. I cannot take anything bad you say about onboard seriously, when you show such a prejudice. Without evidence.
justinb885 wrote:Erssa wrote:You cannot just pick your HDs and stick them to a compeletely different controller and expect them to work there. Poorly implemented drivers with hardware controller can lead to data loss just as much in hardware controller then in onboard. There is absolutely no reason to speculate on the "poor implementation" of onboard controllers drivers, until even one concrete shred of empirical evidence of drivers causing data loss is presented. To be considered valid this evidence will have to be repeatable in test environment.
I find that amusing.
And yet you will willfully plug your drives into a fake trick BIOS RAID implementation without any such information?
Yes. Innocent until proven guilty.
Or do you have information for the mainboard in question? Feel free to post it here, then, I'd love to see it.
Asus A8N-VM CSM. Information about the motherboard can be found on
www.asus.com. If I had RAID 5 now, I had no fear of swithching the motherboard to a new one using the same controller.
I realize this is SPCR and people are mostly interested in quiet systems and have little interest in the realities of why things work, but I see now reason for you to wander around with a badge of ignorance on your shoulder while you contribute to burning other people.
Bad show.
I consider onboard RAID 5 just as good as onboard RAID 0 and RAID 1, that so many users have. Again I challenge you to give us (SPCR) the evidence behind the conspiracy of manufacturers, implementing data destroying onboard RAID 5 options to their motherboards.
In either case, dispelling continued misinformation to the two people that might be reading this thread besides you and myself isn't worth the effort.
Yes, it's a hard task, but I don't mind the effort. I dispelled the misinformation you presented about stripes and the misleading information regarding "final drive".
jasonb885 wrote:I believe the poster already had a solid idea of what solution was going to be implemented.
The purpose of the poll is unknown.
Erssa wrote:I never looked for a definitive answer to a question with this poll. It's purpose was that of any normal poll. Finding out where people stand and what are their preferences.
I just wonder how many times I have to repeat or quote myself. Is it reading comprehension, or is the consept too hard?
Besides, having a poll with such complicated subject matter behind each option defies a concrete lack of understanding on behalf of the OP.
Adult people have the right to vote for a president in presidential elections, even though most of the voters have no idea what kind of effects and consequences their choises have. This is the reason why I personally object to the idea of people voting for things such as EU constitution, when most of them have no idea, on what kind of consequences the results will have. However this is just a poll. No matter how much or how little you know, you have the right to your opinion. It seems that you cannot craps this. It is you who's understanding is compromised, if you cannot grasp the purpose of polls in general.
Feel free to respond and prove my points wrong. I like the chances to practice my english skills. Civil debating never harms anyone. Although I understand, if you didn't feel like responding, since it seems hard for you to admit it, when you are wrong.