Article: A Beginner's Guide to RAID (Part
1):
Date: Sunday, September 22, 2002
Author: Jsgolfman
Provided By: None
Page: 1 of 1
The Hard Drive
I think we can safely assume that the majority
of our readers have a basic working knowledge of their PC. If
not, let me acquaint you with the item with which RAID is concerned.
This would be your hard disk drive or HDD. Now your HDD is where
all of the software (operating system, spreadsheets, games, etc.)
is stored. Without a HDD, you don't have a PC. Considering the
ever-increasing size of HDD's today, pushing 320GB, this means
a lot of data to be stored. I think we can also assume that you
would like this data to remain on your HDD for an extended period
of time.
As I'm sure you are aware, more and more data
on these disks means, progressively, slower and slower read/write
times. For example, if I click "save" on this document,
this is "writing" to the HDD. If I then close it and
open it again, this is "reading" from the HDD. The time
it takes to put the information on the disk or open the information
from the file is the read and/or write time. As you place more
and more data on the disk, space has to be allocated for each
item. Unless you regularly defragment your disk and practice proper
maintenance, the data can be placed at various locations on the
disk. I won't go into the actual description of HDD, if you are
curious you can look here. For our purposes, we'll skip the HDD
101.
RAID
No, it's not the can under the sink that
Kills Bugs Dead. RAID is an acronym for either: Redundant Array
of Inexpensive Disks or Redundant Array of Independent Disks.
The concept of RAID is not a new one, at least
in technology terms. It was developed by the computer science
department at UC Berkeley back in the 80's. That's where the Inexpensive
part came from. The term Independent is now in use, but the concept
is the same. Fine, you say, but what does all that mean? Well,
let's look at the operative words in the phrase, Redundant and
Array. If I say, "That politician is a crook!" I am
being redundant, or possibly naïve. I am repeating myself.
This is an accurate portrayal of the meaning in this context,
for the most part. As you'll see, you can have a RAID array without
the redundancy part. Speaking of an Array, what we mean here is
two or more HDD's grouped together that appear as a single HDD
to your CPU. How this is accomplished will be discussed later
in Part 1. Depending on where you look you can find a multitude
of RAID levels. We'll be dissecting three of the most common in
this section. They are level 0, level 1 and level 0+1, sometimes
referred to as 10 or 0/1.
You may often see RAID 5 used also. There are
technical differences which are, frankly too complex to discuss
in this short article. For our purposes, we shall use 0+1 and
consider this "striping with parity".
At the same time there are two distinct
types of RAID. These are software and hardware. Which type of
controller you use makes the designation. With hardware RAID control
is accomplished by a PCI card, similar to a sound card or through
the use of a motherboard with a RAID controller. Software RAID
performs this same task as part of the operating system. High-end
motherboards today often come equipped with on-board RAID controllers.
Your choice of motherboard will often determine which RAID type
you choose. The two most common and popular are Promise and High
Point. This decision rests with you. I won't recommend one over
the other. Weigh the benefits and costs, and then choose your
weapon.
Operation
By now we've firmly established the definition
of RAID and where in your PC it may come in handy. At this point
you are most likely asking, "Yeah, but how does it work?"
Thanks for asking; I'd be happy to explain.
As I mentioned above, there are 3 common levels
of RAID we plan to focus on. RAID 0, RAID 1 and RAID 0+1. It's
not essential that we decide on a type just yet, so let's leave
that till later. To get things going smoothly, how about we start
with the simplest first?
RAID 1
RAID 1, also known as "mirroring",
does just that. This is where it earns its title of redundant.
In this setup all the data is written to each disk simultaneously,
in effect "mirroring" the data to each disk. I know,
a mirror image is backwards, but you get the idea. This is the
optimal array if system integrity and data backup is the goal.
Your CPU will only see one large logical disk even though you
have two or more physical disks.
Remember the read/write speed I mentioned
earlier? Well, this is the drawback of using RAID 1. Since the
same data must be written on multiple disks, these times might
increase. You may be lucky and they won't, but don't look for
any increases either. However, if you are just looking for a backup
you could try a tape drive, zip/jazz drive (for smaller needs)
or one of a multitude of software programs. If one of those options
will suffice or this doesn't sound like something you need, why
not move on to the next level, RAID 0.
RAID 0
RAID 0, also known as "striping",
is the process of "writing" the data across two or more
disks simultaneously. It sounds similar to RAID 1, except that
it doesn't write all of the data to each disk. Striping involves
writing portions of the data to each disk. For example, on a two
disk RAID 0 array, a section of data will be written to disk 1,
another section of data from the same file will be written to
disk 2, section 3 will be written to disk 1, section 4 will be
written to disk 2 and so on. All of this writing to disks takes
place at the same time, just as in RAID 1 except there is no redundancy.
There you have the drawback to RAID 0, no data integrity. If one
disk fails, the whole array fails because the data is spread between
all of the disks. This is a major concern for those of us who
don't use another method of backing up our data. This is a very
real option for individuals who have a machine they use for gaming
exclusively and want the maximum speed during a marathon fragfest
of UT (Unreal Tournament). Gee, I haven't given you any great
reasons to start a RAID array yet, have I? Well, let's take a
gander at another level that might meet your needs, RAID 0+1(0/1,
10)
RAID 0+1
RAID 0+1, also known as RAID 0+1, is the
best of both worlds. You get redundancy and the parity (data protection)
of RAID 1 and the speed of RAID 0. In order to do this you need
a controller that can combine two RAID arrays into a single array
or parity group. So essentially you set up your RAID 1 and RAID
0 and they are in turn set up as RAID 0+1. You get high performance
combined with data integrity. Well, why didn't I just start with
that one and finish there? Because, you need to know what you
are doing before you open that case up and I couldn't have written
this nice guide. But, since you are all fired up about this we
will proceed to your next hurdle. What controller do I get?
RAID Types
In order to achieve all the benefits of using
RAID, you need a controller. A controller manages the array and
transfers of data between the disks. This is the most important
part of the array so don't skimp, find a good one that does what
you want. There are two main types, hardware and software. There
are advantages and disadvantages to each. Software RAID is just
that, software that controls the interaction of the drives and
transfer of data. It can generally handle all of the operations
needed for running a good RAID setup and is generally easier to
set up. Of course, being software based all the functions are
handled by your CPU, it can severely tax its power and adversely
affect its other activities, and it is also usually not as flexible.
Hardware RAID takes this control away from the processor and relieves
it of this burden. It can either be incorporated into the motherboard
or come in the form of an add-on PCI card. The price can range
from $15-$20 more on the motherboard price to $100 or more for
the card. All the functions handled by the CPU in the software
RAID are now performed by the PCI card or motherboard controller,
thus allowing the CPU to work on more important items. Like alt.binaries.whatever.
It also offers more robust fault-tolerant features and increased
performance. There is the question of disabling if you already
own an OS with software RAID and want to switch to hardware. Again,
get one that does what you need at an acceptable speed and fits
in your budget. We won't even get into SCSI.
Conclusion
If you haven't already decided on RAID,
do so. It's a small investment that can make a world of difference.
I covered a lot of ground and moved pretty quickly today. If you
have any further questions, google, there are a million other
amateur Ph.D's out there and I'm sure you can find one that helps.
Thus concludes Part 1 of the Beginner's Guide to RAID. Be sure
to tune in for the next exciting chapter, Part 2: Installation
and Operation of RAID, also known as "This program has performed
an illegal operation and will be shut down".
|
