RAID's & Disk Management

[Volenar]

Back-up strategies

A back-up is a copy of data which are located on the data-carrier. The copy gets created to prevent the loss of data when your computer crashes or gets stolen, ..

When you make a back-up often it becomes a part of the recovery process. When the original data gets lost/deleted/.. the back-up can be used.

Some causes which can lead to the loss of data :

• The data can be damaged or removed.
  
• The driver can break.
  
• The driver/computer can be stolen
  
• ..
  

You can use several back-up strategies, all depending on what causes the loss. It can be that, in your most recent back-up, data was already damaged to due an unknown factor. That is the main reason why it is smart to save more than one back-up from a longer period. When you have to restore the data manually it can take up to months or even years. Taking a back-up only takes a few minutes.

The Different back-up methods.

There are 2 different back-up methods (globally known, there are more but..)

• A back-up of the full hard drive. (image back-up)
  
• A back-up of all data.
  

The image back-up gets used to repair the hard drive in case of a lethal error. The back-up of data gets used to replace the data in case of an error in these files. The data back-up hasn't got any effect on the system, also the data back-up gives you the chance to compare older versions of files with newer versions of files and sort out their pros/cons, without it all being on saved on the hard drive.


The Different back-up types.

The copying of files.
A simple form of back-up is when you just copy an important file to a separate spot on your hard-drive/back-up station. This can be very effective on some cases.

A full back-up.
A copy of all data on the hard drive gets created, file by file. Due to the time that it takes to create this kind of back-up there will be a time span between the creation and the point on which the back-up actually works due to the fact that it backs all files on the hard drive up. Meaning that ; the more files you have on your hard drive the longer this back-up will take. This back-up gets saved for a long time mostly even if there is a newer version of it.

Incremental back-up.
The incremental back-up is a back-up which only backs all files up which got changed since the last back-up. This is the so called growth strategy.

The incremental back-up requires the least data space, but it'll ask for more time and back-up media in case of data loss. Mostly a weekly full back-up gets created and the other days they take an incremental back-up.

To go back in time, this is a classic example of an incremental back-up command in MS-DOS :

Code:

xcopy c:\source\*.* d:\destination\*.* /s /m

Differential back-up.
A differential back-up takes a back-up of in cycles. The changes made the other days since the last full back-up will be saved in a different back-up copy. When the full back-up has been applied the differential back-up can be removed.

Progressive back-up.
The progressive back-up is similar to the incremental back-up and gets called "incremental forever back-up" from time to time. This similar item requires a back-up server that can save back-up copies.

The first back-up created is a full back-up and from that point being only incremental back-ups get created. Whenever it is necessarily to restore a full file-map to its original position there is no need for several back-ups. The back-up server builds a list of all the files which are required and sends them back to their original position.

Raid & Disk Striping

The term RAID (Redundant Array of Independent Disks) and/or Disk Striping supports on the principle of data being maintained on a safer way on multiple hard drives. This allows quicker access to the disks.

A hardware RAID exists out of multiple hard disks and a RAID controller. The principle is that the data gets maintained on multiple hard disks. A RAID is considered as one single logical hard drive. RAID uses the technique striping which splits up the available space of each hard drive into separate parts of a certain size.

There are 2 sorts of disk striping : Single-user and multi-user.

• With single user there can be multiple disks processing multiple I/O tasks coming from the same work-station.
  
• With multi-user disk striping multiple I/O tasks coming from multiple works-stations at a time can be processed by multiple hard drives. This means that when one hard drive is processing a read/write task coming from one works-station a other hard drive can process an other read/write task coming from another work-station.
  

Some vocabulary :

• RAID : Redundant Array of Independent Disks
  
• Striping : Splitting the size of a disk in equal parts.
  
• MTBF : Mean time between failure.
  
• JBOD : Just a bunch of disks.
  

RAID-types

RAID 0
RAID 0, also known as disk striping is the process with which a certain amount of data gets divided into 2 blocks. These blocks are spread over separate partitions on at least 2 hard drives. Each stripe has the size of the smallest partition.

When you have e.g. 3 partitions selected 150 MB each, an other one being 100 MB and a fifth partition of 50 MB. Each stripe will have a size of 50 MB in this case.

Its a mandatory to make all partitions the same size in order to lose no memory. Each created stripe is a part of the stripe-set. The use of RAID 0 results in a speed increase because the data is split onto several drives, allowing parallel processing.



RAID 1
RAID 1 also known as disk mirroring exists out of at least 2 hard drives who are an exact copy of each other, but there is not striping. The 2 disks get viewed at as one.

If one of the disks fades the system won't notice this and it will continue working the way it used to. There will be sent out a signal that there is something wrong, of course. The performance to read is very good, knowing that 2 individual hard drives can process a task. The disk speed is the same as with the use of one single hard drive.

RAID 1 is one of the less effective methods to maintain data safely and it might be very expensive as well. It is a simple yet loyal solution when it is about security.



RAID 0 +1
This combines the techniques of the previous 2 types. In order to use this you need at least 4 disks/partitions.



The big advantage of this RAID-type is the quick, fault-tolerant system.

• The data can be saved on a safe way.
  
• The data can be read fast, which means double speed.
  

This RAID-type knows some disadvantages as well.

• You need at least 4 hard drives.
  
• Because everything gets saved twice, only 1/2 of the available disk space gets used.
  
• The written happens slower due to the fact that everything gets written twice.
  

RAID 5
RAID 5 uses at least 3 hard drives. This means that the loyalty increases yet you use your drives efficiently. No exact copy gets created but the data gets divided over all the hard drives. You maintain some extra information (the parity) with which you can reconstruct data if one of the drives breaks.

The use of RAID 5 has a big advantage : the data is safe + it doesn't need to be saved twice.

The disadvantage of RAID 5 is that you require at least 3 hard drives. The writing of data costs some time as well due to the fact that the parity needs to be changed each time.

The parity is also known as "Check Sum"




Thank you for reading this guide.

Kind regards, Volenar.

[Telhast]

I've learned a lot after reading this thread . Was confused between RAID 0 at first . But after reading the RAID 0 part you wrote , I know I've mastered my RAID's . :-D

[Lulz]

Nice share!
You are very intelligent!