Security to acting like a sled, the
Security can mean many things.
In the computer world it refers to how safe your data is. For obvious reasons this is of the utmost importance to the business world. A loss of data means a loss of man-hours and a frantic search for hard copy.
Most, if not all businesses have hard-to-recreate documents and data that is essential.In the worst case scenario a business will come to a screeching halt and the losses can be catastrophic. In order to keep this from happening various methods have been developed to back up or make copies of data.
In 1987 computers were wide spread in large companies. These companies made backup tapes to insure that their data was safe, but it was obvious that there was a need to make improvements. A hard drive death meant that at lest some new data was lost.
The search for fault tolerance in computers led to the development of RAID. A group of computer engineers at Berkeley published a paper entitled ” A Case for Redundant Arrays of Inexpensive Disks”. Hence the acronym RAID. The basic idea was to combine multiple small, inexpensive disk drives into an array of disk drives which yields performance exceeding that of a Single Large Expensive Drive (SLED) .
In addition to acting like a sled, the RAID appears to the computer as a single logical storage unit or drive. Five types of array architecture are defined in the Berkeley paper, RAID-1 through RAID-5. The new term RAID-0 is used to refer to a non-redundant array of disk drives.
Each provides disk fault tolerance by storing information in various ways. The rest of this paper is going to describe the different RAID levels list the pros and cons of each and describe which RAID levels are good for which applications . This is a representation of a single hard drive. Blocks DO, D1, D2 are blocks of data. When they are read from the hard drive they are read out in order. This form of data storage offers no fault tolerance.
In other words if your hard drive dies in a fiery crash you will lose all the data on you hard drive. Since several independent sources estimate that the amount of data storage is increasing by 50 to 100% per yearplanning for disaster recovery and prevention is more important than ever. RAID levels offer fault tolerance, which simply means the capability of a system to handle disasters in such a way that continuous operation is maintained. RAID can downgrade a potential disaster to a mild annoyance.
Raid 0 splits the data across drives. This level is commonly referred to as striping. Striping means that data is read and written in parallel.
The disks fill up equally with information. Since the data is dispersed among several drives your computer has a higher data throughput rate. The simultaneous acquisition of data from all drives makes for a fast retrieval of information. Striping can be used to enhance performance in either a request rate intensive or transfer rate intensive environment.
Also as the number of drives increases, so does the speed of the array. Unfortunately the faster it gets the more likely you are to lose data. RAID Level-0 is not redundant. There is no data backup so if your hard drive crashes you will lose all the information stored on that drive.
High Performance, Highest level of fault toleranceNo cost penalty – all storage is usabledisk failure will cause the entire an entire array to be inaccessibleCost, need twice the data capacity the application requiresTypically smaller capacities of mission critical data.RAID Level-1 disks are mirrored. This means that writes to drive Do are written at the same time to the other Do disk, so you have an exact copy of all data written. If one hard drive fails no information will be lost. Fault tolerance is achieved through redundancy. The performance on RAID Level-1 tends to be faster on reads and slower on writes compared to a single drive.
Faster writes can be achieved if both disks in a mirror pair are on separate I/O busses, this is known as duplexing. RAID 1 is very popular but also expensive because half of the drive space is committed to fault tolerance.Higher read performance that a single diskDisk Striping with Error Correction CodeRaid Level-2 is intended for use with drives which do not have built in error detection. All SCSI drives support built-in error detection, so this level is of little use when using SCSI drives. RAID 2 uses an error checking and correction (ECC) technique widely employed with random access memory(RAM). ECC codes are called hamming codes. Generating hamming codes, the controller disperses data at bit level on multiple hard drives.
The picture shows that 8 bits of data are stored onto 8 hard drives, at the same time, 4 bits of hamming code are generated and stored in 4 hard drives. It is possible to read or write onto all 12 hard drives simultaneously or in parralell. If any of the Hard drives fail, the redundancy scheme can restore the original data in real-time, so threat the system operation is not interrupted. RAID 2 has become outdated do to recent improvements. Commercial products rarely employ RAID 2 Error checking and correction of RAMHard drives integrate strong error checking and correction functions today, so the error checking is not necessaryThis picture was inserted to explain parity.
Parity is a disk storage configuration where additional data is written in separate drives, either with bits (RAID-3) or in 64K blocks (RAID-4) so that in the event of a single drive failure, all data can be reconstructed. These redundant segments of data are used to provide fault tolerance for stored material. As you may have guessed parity is the replacement for error checking used in RAID level 2. Parity checks the data by adding a mathematical formula to the data, where the number of 1s in the data must always be the same.
This data-checking system is known as parity checking. It allows the system to check to ensure the data is correct and uses less disk space than ECC or hamming code. RAID Level 3 uses striping and parity. It is a simplified version of RAID-2. Instead of multiple ECC bits it uses bit parity. RAID-3 disperses data at the bit level, as in RAID-2. There is an array of hard drives for data and one for parity.
Parity is derived from bit 0 through bit 7. If any of the hard drives fail, it restores the original data by and XOR (exclusive-or) between the redundant bits on other hard drives and the parity hard drive. With RAID 3, all the hard drives operate constantly. Data chunks are much smaller than the average I/O size and the disk spindles are synchronized to enhance throughput in transfer rate intensive environments. This makes RAID Level 3 a good choice for networks that use CAD or graphic imaging type applications.
Only one disk is used to store parity data, if this disk fails, the data may not be recoverable depending on the last backup. High performance for transfer rate intensive applicationsCost effective – only 1 extra disk is required for parityCan satisfy only 1 I/O request at a timePoor small, random I/O performanceIf the parity disk fails, data may not be recoveredLarge I/O request size applicationsDisk Striping with large blocks and Parity DriveRAID level 4 stripes data in 64k blocks across several drives, with parity stored on one drive. Data is written in entire blocks on each disk. Information is accessed on a “when needed basis”.
This means that the hard drives in the array can operate independently. RAID Level 4 striping uses parity information to ensure the data is written correctly. The parity information is stored on one disk.
As a result, each time data is written to the stripe set, the parity must be calculated and written to the parity disk. This system works well with the large data blocks, but is not efficient because the parity information must be written as the data is written. If one hard drive fails the parity drive allows recovery of datacost is fairly low, since only one drive is used for parityIn Large systems, requires significantly less disk space than Mirroring. processing time longer than RAID-3writing data, takes longer due to retrieval time of old data and parity information and then written with renewed data and parityFixed Parity disk causes Bottlenecks especially for many small I/O operations scattered randomly and widely scattered across the disk in the arrayBest suited for applications such as graphics, imaging, or video that require reading and writing Huge, sequential blocks of Data.
Disk Striping in blocks and Parity Drive (dispersed)This picture is good but I have a better one that, in my opinion, explains the concept better. I had to leave this one in for the sake of consistancy.RAID Level 5 solves a problem RAID level 4 has. In RAID Level 4 the parity information is stored all on one disk, this can lead to bottlenecks. In RAID Level 5, the parity is dispersed and stored in all hard drives. There are no dedicated parity drives and additional drives are easy to add to the array.
From the Windows NT operating system perspective, RAID-5 is seen as a single volume of large size. Data is striped from storage device to storage device in sequence and by sectors. Fata chunks are much larger than the average I/O size. Disks are able to satisfy requests independently which provides high read performance in a request rate intensive environment.
Windows NT thinks that it is reading data from successive sectors but the sectors are actually on sequential Drives. Each write request requires 4 independent disk accesses to be completed. First old data and parity are read off separate disk. Next the new parity is calculated and finally the new data and parity are written to separate disks. This makes the write performance of RAID Level 5 poor.
Write caching is one way to compensate for this. Parity information is created by comparing the data from one disk to the data on the others one bit at a time. This comparison is done across all the data disks of the array and uses the “Exclusive Or” Boolean function. The Boolean function refers to the “on—off” 1 or 0 values given to parity bits. This means that every bit on the array can be determined by simple algebra.
If you add up all of the stream bits from disk 1, disk 2 and disk 3, in binary, you get their values. Now if you remove a disk you can regenerate the missing data by replacing for the value of X. All this really means is that if a disk goes bad the information from the other disks can be used to replace that information when you replace the bad drive. Some hardware for RAID does this automatically whenever the controller detects a drive failure. This is called having a hot spare.
RAID Level 5 can survive one disk failure, but not two or more.RAID Level 5 is the most common RAID level found on modern sever systems. It is considered to be a convenient compromise between performance, cost and fault tolerances.poor write performance can be compensated forNo performance gain in data transfer rate intensive applicationsIdeally suited for applications with many, small I/O operationsBibliography: