Raid levels...can someone explain them for me

Norbie

Can anyone explain the RAID levels to me...or give me the link to a really good tutorial?

Mishra

Honest question...

What is google providing for you that doesn't answer this question?

Zaits

Let me google that for you

Plantwiz

Norbie wrote: »

Can anyone explain the RAID levels to me...or give me the link to a really good tutorial?

What specifically do you want to know?

Are you looking for definitions?
Are you looking for how raid works?
Something else?

RAID Level Explained
RAID Levels Explained

And as metioned above, simply 'google' RAID and see what make sense to you.

earweed

There are many different RAID levels and most people don't remember all of them anyway unless they are constantly responsible for different types of RAID. The big ones to know, at least for Microsoft, are RAID 0, 1, and 5. There are also the 0+1 and 1+0 RAIDS that you should be able to distinguish from each other.

za3bour

Here is a tip I always use when I want to search about something I don't know I use Google advance search to look for doc/pdf files that are usually free.

So if you did try to search for PDF files about RAID you will find a lot of resources from companies, universities, cources ..etc

So Google first and then refine what you really don't understand or need extra info about then post it here and many will help you.

anobomski

This is one of the best on the web:

AC&NC | RAID.edu - RAID Levels - RAID Level 0 - RAID 0

Norbie

Mishra wrote: »

Honest question...

What is google providing for you that doesn't answer this question?

I would like someone to actually explain it for me in a way that clarifies for it me and maybe I can bounce questions off them and maybe have some sort of discussion. But if that's too big of an imposition for you or the other users then mea culpa. My goal was that when I take the server+ test again maybe I'd have some good tips on how to calculate some of this stuff.

I don't get why I get punked out for asking a question when these forums are filled with them.

A part of me thinks there is a fair amount of derision on the server+ board. If someone asks a question on this board it tends to be followed with a snarky response of "don't worry about learning that...study for the microsoft server exams". Which is a fine point to make but maybe some people just want their question answered instead of/in addition to comments from the peanut gallery.

On the other boards my experience has been that everyone puts their best foot forward and is friendly and supportive even to the newbies who ask simple question that yes even a google search could answer.

Sorry to have wasted all of your time.

earweed

You asked a VERY broad question. What is it you don't understand?

rsutton

I recommend reading this: How to ask Questions the smart way. Have you done your own research? If you have, it is not apparent in your question.

Mishra

Asking a question is fine even if isn't a "smart" way...

But if I come back with a question, answering it will probably help me in responding to your request.

I said "Honest question..." for a reason.

Norbie

earweed wrote: »

You asked a VERY broad question. What is it you don't understand?

A LOT of it. But for starters:

The differences between say Raid 0+1,1+0, 5+0, 0+5, 1+5 and 5+1.

Assume I am a newbie who doesn't already have a job dealing with servers but that someday that is my goal.

If I can get a detailed response from someone...I thank you for your assistance. But if it's a response to mock or bait me like the gentleman who asked if I knew how to ask questions the smart way and if I did any research on the subject just please don't bother. I wanted someone to help me connect some of the dots I couldn't articulate because I honestly DON'T know enough about the subject matter to properly formulate the questions needed to get the information I need. Right now it's all abstract (since I don't work with raid) and I was hoping someone could give me some sort of clarity on the subject. I did do a google search prior to my original post, I have a server+ book and
a transcender cd and the practice questions from ucertify...but if you give me a problem like on the server+ where I have to decide between say a RAID 5 array of 20 drives and Raid 1+0 of 10 drives I don't how to look at that problem and say hey's that the right answer.

Plantwiz

Norbie wrote: »

A LOT of it. But for starters:

The differences between say Raid 0+1,1+0, 5+0, 0+5, 1+5 and 5+1.

Assume I am a newbie who doesn't already have a job dealing with servers but that someday that is my goal.

If I can get a detailed response from someone...I thank you for your assistance. ....

Did you bother to read my response and recommended websites?

From those sites, what do you not understand? I know for certain one clearly explains the RAID types and gives examples. Beyond that, I'm not clear what of that doesn't make sense for you? You may want to invest a little time and setup a RAID because it sounds as though you may be lacking experience which should clear up much of the confusion. Aside from this, there is hardware raid and software raid and generally in servers one handles hardware raid for the benefits of hardware raid over software.

I did do a google search prior to my original post, I have a server+ book and
a transcender cd and the practice questions from ucertify...but if you give me a problem like on the server+ where I have to decide between say a RAID 5 array of 20 drives and Raid 1+0 of 10 drives I don't how to look at that problem and say hey's that the right answer.

Ok...

start with this... take your above scenario and answer for us what is a RAID 5 and what is a RAID 1+0. Don't worry about the drive counts...just describe RAID 5 and RAID 1+0.



As far as you feeling that members are a little snarky with you, you did ask a very general question and frankly without providing more detail to your question, I don't have a way to answer your question without asking you for more detail (which another member did as well). In return, you respond that you don't think anyone is helping you Ok....well re-read your server+ book (and whose book is it? Sybex?) You may toss your practice questions aside. They don't help teach a candidate anything. Grab the objectives and go point-by-point to research this material. If you are lacking on experience, then you may need to set-up a lab to get some practice and you will need to do a bit more legwork on the research side.

Your task to prep for this exam will not be impossible, but the exam is geared toward a candidate WITH experience so you'll need a bit more research to get there if you are electing to go about it without actual experience.

Though, without ANY live RAID experience, I am curious why you have elected to take the SERVER+ exam at this stage? Again, it is doable, but not a typical exam and generally many techs skip in favor of exams that will improve their resume.

I think you need to get a RAID card, some drives and just practice. RAID is not difficult. You need to memorize the basics. Set up a couple raids and thats about it. RAID is not much more then how the data is spanned acrossed two or more drives (and or controllers). One typically selects the raid type based on the type of server, the amount of risk to the data they can afford, and the type of backup...though this should be covered in your Server+ text. As far as which is best? It depends....

Mishra

There are reasons why explaining what RAID is isn't necessarily an easy task for us. That's why there are large articles out there discussing what RAID is. Many pages of information talking about what it is or what it does.

Now, I know that sometimes I like to ask general questions to simply get a base understanding so I can go out and research more. Like on wikipedia the general definition of RAID is this

"RAID, an acronym for redundant array of independent (or inexpensive) disks is a technology that provides increased storage reliability through redundancy, combining multiple relatively low-cost, less-reliable disk drives components into a logical unit where all drives in the array are interdependent."

Honestly, I think this is a terrible example if I'm trying to explain RAID to a 9 year old. I would say something more like this.

"RAID is a method to take more than 1 hard drive and provide yourself a way to increase performance or save yourself from 1 hard drive failing and losing all your data."

I prefer my explanation much better. I can take that information and read this "RAID 0 helps to increase performance by striping" and apply it to the base definition of what I said much better than the first.

However, there are many articles out there explaining RAID. And the question I proposed was because the skill to search Google and understand articles is much better than asking in a forum. Seriously. It's a skill and training to search the internet to get the responses you want to help you understand ANY subject. If you read RAID articles like you said you did, and didn't understand something, I need to understand what to better help you find the information you need.

genXrcist

Norbie wrote: »

A LOT of it. But for starters:

The differences between say Raid 0+1,1+0, 5+0, 0+5, 1+5 and 5+1.

Assume I am a newbie who doesn't already have a job dealing with servers but that someday that is my goal.

If I can get a detailed response from someone...I thank you for your assistance. But if it's a response to mock or bait me like the gentleman who asked if I knew how to ask questions the smart way and if I did any research on the subject just please don't bother. I wanted someone to help me connect some of the dots I couldn't articulate because I honestly DON'T know enough about the subject matter to properly formulate the questions needed to get the information I need. Right now it's all abstract (since I don't work with raid) and I was hoping someone could give me some sort of clarity on the subject. I did do a google search prior to my original post, I have a server+ book and
a transcender cd and the practice questions from ucertify...but if you give me a problem like on the server+ where I have to decide between say a RAID 5 array of 20 drives and Raid 1+0 of 10 drives I don't how to look at that problem and say hey's that the right answer.

Not trying to be a smart @$$ but in the IT world if you ask general questions that you can get answers to yourself, you'll annoy others and get snippy responses. Asking a general question requires a lengthy answer as the expert doesn't know what they're trying to answer. Why should the expert put 15 minutes into an answer that took you 5 seconds to post? Pretty annoying imo. I like answering and explaining concepts to newbies who show me a desire to learn. Short and vague questions show the opposite.

You just need to understand the following terms:

Spanning - Drives combined for the purpose of increasing storage volume. Offers no performance or redundancy value.

Mirror - At least Two physical or logical (virtual) drives that "mirror" each other so the same data is written to both simultaneously in order to protect you in the event that one HDD failed. Offers no Performance value but does offer redundancy at the cost of lost storage (i.e. the mirror'd drive)

Striping - Data is broken up and written to three or more drives for the purpose of increasing performance. It does this by writing the broken data to seperate drives simultaneously. (***I heard Da Vinci could do this by writing with both his left and right hads at the same time! COOL!***). Performance gained, storage not impacted but at the cost of no redundancy.

Parity - Based on Boolean logic and it would take too long to write up a proper definition so suffice it to say that it allows for Drive failure because with parity, missing data that had been written in a 'Striped' RAID solution can be recovered with Parity values that remain in the other Drives. Remember it like this. D=Disk | W=Write P=Parity
***3 HDD RAID 5 (Striping with Parity) setup writes data like this:
1st Data write sweep ---> D1(W) - D2(W) - D3(P)
2nd Data write sweep ---> D1(W) - D2(P) - D3(W)
3rd Data write sweep ---> D1(P) - D2(W) - D3(W)
With this, storage is lost (equivalent of 1 drive in the array), Read performance is improved and Write goes down but most importantly, redundancy is gained as the Parity allows data to be rebuilt.

So after that, why don't you try and explain to us what 0(Striping) + 1(Mirror) means vs. 1+0?

Plantwiz

genXrcist wrote: »

....

So after that, why don't you try and explain to us what 0(Striping) + 1(Mirror) means vs. 1+0?

I asked the OP this and another question and have yet to see a response. Others have also asked questions.

So, we shall see, but while the OP criticizes the members here for their responses, I didn't see anyone yet recommend the OP try the search feature to review the archive of posts here where RAID has previously been discussed to see if any of those posts assist in clearing up the confusion.

However, this would not be the first time a new member posts a question which needs further clarifciation to be answered properly, complains about being asked to clarify and doesn't return.

No one is being 'mean' we need to know what you don't get, but in that process the student (that's you) needs to put forth a bit of effort and get the terms down and try to setup a RAID to better understand what you don't know.

genXrcist gave a nice simple breakdown of the terminology. However, we have some RAID experts around who'll talk your ear off, but as genXrcist mentioned it will be helpful if you put more then a few seconds into your question to get a lengthy answer.

And please do tell us the text you are using!

genXrcist

Plantwiz wrote: »

As far as you feeling that members are a little snarky with you, you did ask a very general question and frankly without providing more detail to your question, I don't have a way to answer your question without asking you for more detail (which another member did as well).

I like the term 'Snarky'. LOL Gonna use that going forward.

Devilsbane

genXrcist wrote: »

I like the term 'Snarky'. LOL Gonna use that going forward.

Urban Dictionary: snarky

Plantwiz

genXrcist wrote: »

I like the term 'Snarky'. LOL Gonna use that going forward.

I didn't make it up, but feel free to share the love

chdotson2000

I will try to answer quickly but yet still provide the differences between each option. Please excuse any grammar mistakes as I am a Computer Technician and not an English teacher. Hopefully you will understand.
RAID = Redundant Array of Independent Disks or Redundant Array of Inexpensive Disks. I have seen the acronym used both ways. Not sure why it would be “Inexpensive” as in all my dealings with RAID it cost more to use RAID, however the cost may be cheaper then loosing data or system downtime.
Note all drives in a Raid Array must be the same size. If not then the smallest drive size will be utilized on the other larger drives.
RAID 0 (Spanning) = 2 or more disks combined to represent 1 larger logical disk. Example 500 GB x 2 = 1 TB of disk space. This is used to increase performance as that both drives read and write independently of each other increasing seek/read/write times. Data is stored across both drives so a file may actually reside partially on both drives. All data will be lost if the RAID array is separated or in the event 1 or more drive fails. Normally used on Gaming Computers or where speed is most important. If you choose this option I recommend you set your cluster size to its largest size (128 Kb). I think 4kb is the default size using windows NTFS. Since speed is the result you want by choosing this array it is faster and easier to find and pick up a rock in a pile of rocks than it is to find and pick up a grain of sand in a pile of sand.
RAID 1 (Mirroring) = 2 or more disk combined to represent the same disk. Sounds weird when it’s said that way. But it comes down to redundancy (duplicating) without “parity”. Example 500 GB x 2 = 500 GB of disk space. Where did half the space go you say? It is mirroring the first drive in the array. All drives have the exact same data on them. If 1st drive fails the other drive has all the data backed up and can be restored once the faulty disk is replaced. It is useful but not used as much as you might think, as restoring the data from the other drive could be time consuming depending on the amount of data present. You will see why “parity” is the tool of choice when redundancy is the number one concern.
RAID 5 (Spanning with Parity) = 3 or more disk combined to represent 2/3rd of the total drives. I know that sounds bad, but the fault tolerance is what you’re looking for here. Example 1 TB x 3 = 2 TB of disk space. Why would you give up 1/3rd of your drive space? Here is why. Data is spanned across the disk array as in RAID 0, so all data is broken up and stored on all 3 drives. So the file “Walk the Dog” is stored on the drives as such Drive one “Walk”, Drive two “the”, Drive three “Dog”. All drives are needed to open the file. But parity steps in and makes life easy in the event of 1 drive fails. 1/3rd of each drive contains the information of another drive in the array. So Drive 1 “Walk” also has the information of Drive 2 “the”, and Drive 2 “the” also has the information on Drive 3 “Dog”, and Drive 3 “Dog also has the information on Drive 1 “Walk”. So if any but only one drive fails all data is automatically restored when the faulty drive is replaced. These drives are normally hot swappable, meaning you can change the drive out without powering down the system. It is used when system downtime is not an option. With this option the more drives you use the less space you lose to parity storage. So if you use 4 drives then the parity is 1/4 of each drive, or 5 would be 1/5 of each drive.

So back to your more clarified question of wanting to know what “Raid 0+1,1+0, 5+0, 0+5, 1+5 and 5+1” are
These would be combination of the others that I have explained.
RAID 0+1 could be 2 x 300 GB raptor drives making 1 logical 600 GB drive. Then that array would be duplicated with another 2 X 300 GB raptor drive array backing up the first set of drives. So we have 1.2 TB of drives giving us a very fast 600 GB logical drive while a second very fast 600 GB logical drive backing up all data on the first drive array in case 1 of the drive arrays fails. This gives us the best of RAID 0 and RAID 1 at the same time.
RAID 1+0 could be 2 x 500 GB drives making 1 logical 500 GB drive with each containing the exact same data. Then those drives would be spanned to another set of 2 X 500 GB drives each contains the exact same data. In my opinion this array does provide redundancy but doesn’t provide the performance you are looking for. Essentially all drives contain the same information D1 is mirrored onto D2 and then D1 and D2 are spanned to D3 and D4. So D1=D2 with reference to data as well as D3=D4. Also D1 and D2 are in another performance based spanning array with D3 and D4. This brings so many concerns to the table. If the Raid 0 array has a fault the Raid 0 array is pretty much useless. But I could be looking at this from the wrong view point. It had to be useful for something or it wouldn’t have been designed. It also seems like a terrible waste of drive space in the complete array we have 4 x 500 GB = 2 TB of space but only providing us with 500 GB of useable storage.
RAID 5+0 could be 3 x 1 TB drives providing 2 TB of useable space (parity is using the final 1 TB of space) these drive most likely be hot swappable and automatically recover information due to a fault. Then a second 3 x 1 TB drive array would span the data in the first three drives. So once again 6 TB of drives providing 2 TB of useable storage space. But theoretically this would allow 2 drives to fail at the same time as long as each is in their own respective arrays. Not sure in reality if that would work, and there are more efficient ways to provide fault tolerance then this.
RAID 0+5 could be 2 x 300 GB raptor drives making 1 very fast 600 GB logical drive. Then 2 more 2 x 300 GB raptor drives making 2 more very fast 600 GB logical drive arrays. These would span the information across the drive array and make it fault tolerant however, 2 drives would be replaced in the event of even 1 drive failure. For example if drive 3 fails then drive 3 and 4 would be replaced since they were in a Raid 0 array. Okay now the drive usage gets crazy here. In total we have 6 x 300 GB drives having a total 1.8 TB of space. But since parity steps in and take away 1/3 of the drive space for redundancy only 1.2 TB of space would be useable. Still better than RAID 5+0 and Raid 1+0 in my opinion. But again this was designed out of a necessity for it.

Raid 2, Raid 3, or Raid 6 could give you better fault tolerance. But that would be another 1400 words to discuss and compare those. Wikipedia has an informative page with pictures (I love pictures) to show the drive arrays and disk usage as well as parity. http://en.wikipedia.org/wiki/RAID
I hope I have at least hit the ball in the park with reference to your question. And I hope I didn’t confuse you or myself while typing it. But to be honest books get written on this subject and no one post in a forum will answer it all. But I gave it a shot.