CBL Serviços de Recuperação de Dados

Data Recovery Article

Raid Array & Server Glossary Of Computer Terms

-- A --

Array

Multiple disk drives configured to behave as a single, independent disk drive. See also Disk Array.

-- B --

Background Initialization

Where the initialization process of a disk array takes place in the background, allowing use of a disk array within seconds instead of several hours. Also known as Immediate RAID Availability.

Benchmarks

A set of conditions or criteria against which a product or system is measured. Computer trade magazine laboratories frequently test and compare several new computers or computer devices against the same set of application programs, user interactions, and contextual situations. The total context against which all products are measured and compared is referred to as the benchmark. Programs can be specially designed to provide measurements for a particular operating system or application.

Berkeley RAID Levels

A family of disk array protection and mapping techniques described by Garth Gibson, Randy Katz, and David Patterson in papers written while they were performing research into I/O systems at the University of California at Berkeley. There are six Berkeley RAID levels, usually referred to as RAID Level 0 through RAID Level 5. See also RAID Levels.

Bridge RAID Controller

A device appearing as a single ID in a Storage Area Network (SAN), but which bridges to multiple devices, typically used to control external RAID subsystems (compare to Internal PCI-based RAID subsystems, see Internal RAID Controller). A bridge RAID controller is often referred to as an External RAID Controller.

-- C --

Cache

A temporary storage area for frequently accessed or recently accessed data. Cache is used to speed up data transfer to and from a disk. See also Caching.

Cache Flush

Refers to an operation where all unwritten blocks in a Write-Back Cache are written to the target disk. This operation is necessary before powering down the system.

Cache Line Size

Represents the size of the data "chunk" that will be read or written at one time, and is set in conjunction with stripe size. Under RAID EzAssistTM, the cache line size (also known as Segment Size) should be based on the stripe size you selected. The default segment size for Mylex RAID controllers is 8K. See also Stripe Size.

Caching

Allows data to be stored in a pre-designated area of a disk or RAM. Caching speeds up the operation of RAID systems, disk drives, computers and servers, or other peripheral devices. See also Cache.

Conservative Cache

An operating mode in which system drives configured with the Write-Back Caching policy are treated as though they were configured for Write-Through operation and the cache is flushed.

Consistency Check

A process that verifies the integrity of redundant data. A consistency check on a RAID 1 or RAID 0+1 configuration (mirroring) checks if the data on drives and their mirrored pair are exactly the same. For RAID Level 3 or RAID Level 5, a consistency check calculates the parity from the data written on the disk and compares it to the written parity. A consistency check from Mylex utilities such as Global Array ManagerTM (GAM) or RAID EzAssistTM give the user the ability to have a discrepancy reported and corrected. See also Parity Check.

Controller

An adapter card, RAID controller, or other module that interprets and controls signals between a host and a peripheral device.

-- D --

Degraded Mode

A RAID mode used when a component drive has failed.

Disk

A non-volatile, randomly addressable, re-writable data storage device, including rotating magnetic and optical disks as well as solid-state disks or other electronic storage elements.

Disk Array

A collection of disks from one or more commonly accessible disk systems. Disk arrays, also known as RAID, allow disk drives to be used together to improve fault tolerance, performance, or both. Disk arrays are commonly used on servers and are becoming more popular on desktops and workstations. See also Array.

Disk Drive

A device for the electronic digital storage of information.

Disk Failure Detection

A RAID controller automatically detects SCSI disk failures. A monitoring process running on the controller checks, among other things, elapsed time on all commands issued to disks. A time-out causes the disk to be "reset" and the command to be retried. If the command times out again, the controller could take the disk "offline." Mylex DAC960 controllers also monitor SCSI bus parity errors and other potential problems. Any disk with too many errors will also be taken "offline." See also Offline.

Disk Traveling, Drive Traveling

A process that occurs when the drives are placed in a different order than the original order. Disk traveling can occur whether or not a drive has failed.

Disk System

A storage system capable of supporting only disks.

Drive Groups, Drive Packs

A group of individual disk drives (preferably identical) that are logically tied to each other and are addressed as a single unit. In some cases this may be called a drive "pack" when referring to just the physical devices.

All the physical devices in a drive group should have the same size; otherwise, each of the disks in the group will effectively have the capacity of the smallest member. The total size of the drive group will be the size of the smallest disk in the group multiplied by the number of disks in the group. For example, if you have 4 disks of 400MB each and 1 disk of 200MB in a pack, the effective capacity available for use is only 1000MB (5x200), not 1800MB.

Dual Active

A pair of components, such as storage controllers in a failure tolerant storage system, that share a task or set of tasks when both are functioning normally. When one component of the pair fails, the other takes the entire load. Dual active controllers (also called Active/Active controllers) are connected to the same set of devices and provide a combination of higher I/O performance and greater failure tolerance than a single controller.

-- E --

ECC

Error Correcting Code, a method of generating redundant information which can be used to detect and correct errors in stored or transmitted data.

Embedded Storage Controller

An intelligent storage controller that mounts in a host computer's housing and attaches directly to a host's memory bus with no intervening I/O adapter or I/O bus.

External RAID Controller

A RAID controller in its own enclosure, rather than incorporated into a PC or server. External RAID controllers are often referred to as a Bridge RAID Controller. Mylex SANArray FL, FF, FFx, and Pro FF2 controllers are external RAID controllers. Compare with Internal RAID Controller.

-- F --

Failback

Restoring a failed system component's share of a load to a replacement component.

Failover

A mode of operation for failure tolerant systems in which a component has failed and a redundant component has assumed its functions.

Failover Port

A fibre channel port capable of assuming I/O requests for another, failed port on the loop. During normal operation, a failover port may be active or inactive. Failover ports assume the same loop ID and, optionally, the same node from the failed port.

Failure

A detectable physical change in hardware, requiring replacement of the component.

Fault Tolerance, Failure Tolerance

The ability of a system to continue to perform its function even when one of its components has failed. A fault tolerant system requires redundancy in disk drives, power supplies, adapters, controllers, and cabling. Mylex RAID controllers offer high levels of fault tolerance.

Fibre Channel

Technology for transmitting data between computer devices at a data rate of up to 2 Gbps (two billion bits per second), especially suited for connecting computer servers to shared storage devices and for interconnecting storage controllers and drives.

Fibre Channel is expected to replace the Small System Computer Interface (SCSI) as the transmission interface between servers and clustered storage devices. It is also more flexible: devices can be as far as ten kilometers (about six miles) apart. The longer distance requires optical fiber as the physical medium; however, fibre channels also work using coaxial cable and ordinary telephone twisted pair wires.

Flash ROM

Memory on an adapter containing software that can be reprogrammed without removing it from the board.

-- G --

Global Spare

An extra, physical disk drive placed in an array and used as a Hot Spare. A global spare automatically takes the place of a failed drive. See also Auto Swap.

-- H --

Host

Any computer system to which disks are attached and accessible for data storage and I/O.

Hot Replacement of Disks

The design of all Mylex controllers allows for the replacement of failed hard disk drives without interruption of system service. In the event of a SCSI drive failure on a properly configured system (where the data redundancy features of the controller are used), the system generates a message to alert the system operator.

When a replacement drive becomes available, the system operator can remove the failed disk drive, install a new disk drive, and instruct the controller to "rebuild" the data on the new drive, all without interrupting system operations. Once the rebuild is complete, the controller will be brought back into a fault tolerant state. See also Hot Swap.

Hot Spare

A physical disk drive not part of a system drive that the controller can use to automatically rebuild a critical system drive. The hot spare drive must have at least as much capacity as the largest disk drive in the array or the rebuild may not start. See also Hot Standby and Standby Replacement of Disks.

Hot Standby

A redundant component in a fault tolerant storage system that has power applied and is ready to operate, but which does not perform its task as long as the primary component for which it is standing by is functioning properly. See also Hot Replacement of Disks and Hot Spare.

-- I --

Immediate RAID Availability

See Background Initialization

In-Line Terminator

A plug attached to the end of a SCSI cable in order to initiate active termination. Used when SCSI devices on the cable do not have built-in termination. See also Active Termination.

Interface

A hardware or software protocol that manages the exchange of data between the hard disk drive and the computer. The most common interfaces for small computer systems are ATA (also known as IDE) and SCSI.

Internal RAID Controller

A controller circuit board that resides inside a computer or server. An internal RAID controller resides on a bus, such as the PCI bus. Examples of internal RAID controllers include the Mylex AcceleRAID and eXtremeRAID families.

-- J --

JBOD

Just A Bunch of Disks (Drives), a number of disk drives, usually in an enclosure. JBOD implies that the disks do not use RAID technology and function independently. All Mylex RAID controllers support JBOD mode.

-- K --

-- L --

Latency

1. The time between the making of an I/O request and completion of the request's execution. 2. Short for rotational latency, the time between the completion of a seek and the instant of arrival of the first block of data to be transferred at the disk's read/write head.

Logical Drive

The logical devices presented to the operating system. System drives are presented as available disk drives, each with a capacity specified by the Mylex RAID controller. See also Storage Device.

-- M --

Mirrored Cache

A cache memory that has duplicate data from another controller. In the event of failure of the original controller, the second controller can take the cached data and place it on the disk array.

Mirrored Hard Drive

Two hard drives the computer sees as one unit. Information is stored simultaneously on each drive. If one hard disk drive fails, the other contains all of the cached data and the system can continue operating.

Mirroring

The complete duplication of data on one disk drive to another disk drive, this duplication occurs simultaneously with each write operation: each disk will be the mirror image of the other (also known as RAID Level 1, see RAID Levels). All Mylex RAID controllers support mirroring.

-- N --

-- O --

-- P --

Parity

A method of providing complete data redundancy while requiring only a fraction of the storage capacity of mirroring. The data and parity blocks are divided between the disk drives in such a way that if any single disk drive is removed or fails, the data on it can be reconstructed using the data on the remaining disk drives. The parity data may exist on only one disk drive or be distributed between all disk drives in a RAID group. See also Rotated XOR Redundancy.

Parity Check

A function used to verify the integrity of data on a system drive. It verifies that mirror or parity information matches the stored data on the redundant arrays. If the parity block information is inconsistent with the data blocks, the controller corrects the inconsistencies. See also Consistency Check.

Partitioning

Where the full usable storage capacity of a disk or array of disks appears to an operating environment in the form of several virtual disks whose entire capacity approximates that of the underlying disk or array.

PCI Hot Plug

A feature that allows for the printed circuit board (PCB) to be replaced without powering down the entire system-an essential feature in newer PCI-based PCs. Mylex DAC960PG, DAC960PJ, AcceleRAID, and eXtremeRAID products are all PCI Hot Plug compatible. See also Hot Plug.

-- Q --

-- R --

RAID

Redundant Array of Independent Disks, a collection of two or more disks working together in an array. Mylex RAID controllers implement this technology to connect up to 15 SCSI devices per channel. The different forms of RAID implementation are known as "RAID levels." See also Berkeley RAID Levels, Disk Array, and RAID Levels.

The system manager or integrator selects the appropriate RAID level for a system. This decision will be based on which of the following are to be emphasized:

• Disk Capacity
• Data Availability (redundancy or fault tolerance)
• Disk Performance

RAID Adapters

See RAID Controller

RAID Advisory Board (RAB)

An association of companies whose primary intention is to standardize RAID storage systems. Mylex is a member of RAB.

RAID Controller

Low cost RAID controllers that use SCSI channels on the motherboard.

RAID Levels

Mylex disk array controllers support four RAID Advisory Board approved (RAID 0, RAID 1, RAID 3, and RAID 5), two special (RAID 0+1, and JBOD), and three spanned (RAID 10, 30, and 50) RAID levels. All DAC960, AcceleRAID, and eXtremeRAID series controllers support these RAID levels. See also Berkeley RAID Levels.

Level 0:

Provides block "striping" across multiple drives, yielding higher performance than is possible with individual drives. This level does not provide any redundancy.

Level 1:

Drives are paired and mirrored. All data is 100 percent duplicated on a drive of equivalent size.

Level 3:

Data is "striped" across several physical drives. Maintains parity information, which can be used for data recovery.

Level 5:

Data is "striped" across several physical drives. For data redundancy, drives are encoded with rotated XOR redundancy.

Level 0+1:

Combines RAID 0 striping and RAID 1 mirroring. This level provides redundancy through mirroring.

JBOD:

Sometimes referred to as "Just a Bunch of Drives." Each drive is operated independently like a normal disk controller, or drives may be spanned and seen as a single drive. This level does not provide data redundancy.

Level 10:

Combines RAID 0 striping and RAID 1 mirroring spanned across multiple drive groups (super drive group). This level provides redundancy through mirroring and better performance than Level 1 alone.

Level 30:

Data is "striped" across multiple drive groups (super drive group). Maintains parity information, which can be used for data recovery.

Level 50:

Data is "striped" across multiple drive groups (super drive group). For data redundancy, drives are encoded with rotated XOR redundancy.

Note: The host operating system drivers and software utilities remain unchanged regardless of the level of RAID installed. The controller makes the physical configuration and RAID level implementation.

RAID Migration

A feature in RAID subsystems that allows for changing a RAID level to another level without powering down the system.

Read-Ahead Cache

A caching strategy whereby the computer anticipates data and holds it in cache until requested.

Recovery

The process of reconstructing data from a failed disk using data from other drives.

Redundancy

The inclusion of extra components of a given type in a system (beyond those the system requires to carry out its functions).

Rotated XOR Redundancy

XOR refers to the Boolean "Exclusive-OR" operator. Also known as Parity, a method of providing complete data redundancy while requiring only a fraction of the storage capacity of mirroring. In a system configured under RAID 3 or RAID 5 (which require at least three SCSI drives), all data and parity blocks are divided amongst the drives in such a way that if any single drive is removed (or fails), the data on it can be reconstructed using the data on the remaining drives. In any RAID 3 or RAID 5 array, the capacity allocated to redundancy is the equivalent of one drive.

-- S --

SAF-TE

SCSI Accessed Fault-Tolerant Enclosure, an "open" specification designed to provide a comprehensive standardized method to monitor and report status information on the condition of disk drives, power supplies, and cooling systems used in high availability LAN servers and storage subsystems. The specification is independent of hardware I/O cabling, operating systems, server platforms, and RAID implementation because the enclosure itself is treated as simply another device on the SCSI bus. Many other leading server, storage, and RAID controller manufacturers worldwide have endorsed the SAF-TE specification. Products compliant with the SAF-TE specification will reduce the cost of managing storage enclosures, making it easier for a LAN administrator to obtain base-level fault-tolerant alert notification and status information. All Mylex RAID controllers feature SAF-TE.

Sector

The unit in which data is physically stored and protected against errors on a fixed-block architecture disk.

Segment Size

See Cache Line Size

Sequential I/O

A type of read and write operation where entire blocks of data are accessed one after another in sequence, as opposed to randomly.

SES

SCSI Enclosure Services, a standard for SCSI access to services within an enclosure containing one or more SCSI devices. For disk drives, power supplies, cooling elements, and temperature sensors, the actions performed are the same as for SAF-TE monitoring. If a UPS is connected to any SES-monitored enclosures, and an AC failure or two minute warning is reported, conservative cache is enabled and all system drives are switched to write-through cache. Primarily used in fibre enclosures.

Session

The period of time between any two consecutive system shutdowns; system shutdown may be either a power off/on, or a hardware reset.

SMART

Self-Monitoring Analysis and Reporting Technology, the industry standard reliability prediction indicator for both the ATA/IDE (advanced technology attachment/integrated drive electronics) and SCSI hard disk drives. Hard disk drives with SMART offer early warning of some hard disk failures so critical data can be protected.

Spanning

A process that provides the ability to configure multiple drive packs or parts of multiple drive packs. In effect, spanning allows the volume used for data processing to be larger than a single drive. Spanning increases I/O speeds, however, the probability of drive failure increases as more drives are added to a drive pack. Spanned drive packs use striping for data processing. See also Striping and Drive Groups, Drive Packs.

Standard Disk Drive

This term refers to a hard disk drive with SCSI, IDE, or other interface, attached to the host system through a standard disk controller.

Standby Replacement of Disks

See also Hot Spare. One of the most important features the RAID controller provides to achieve automatic, non-stop service with a high degree of fault-tolerance. The controller automatically carries out the rebuild operation when a SCSI disk drive fails and both of the following conditions are true:

• A "standby" SCSI disk drive of identical size is found attached to the same controller;
  
  
• All of the system drives that are dependent on the failed disk are redundant system drives, e.g., RAID 1, RAID 3, RAID 5, and RAID 0+1.
  
  

Note: The standby rebuild will only happen on the same DAC960 controller, never across DAC960 controllers.

During the automatic rebuild process, system activity continues as normal. System performance may degrade slightly during the rebuild process.

To use the standby rebuild feature, you should always maintain a standby SCSI disk in your system. When a disk fails, the standby disk will automatically replace the failed drive and the data will be rebuilt. The system administrator can disconnect and remove the bad disk and replace it with a new disk. The administrator can then make this new disk a standby.

The standby replacement table has a limit of 8 automatic replacements in any session (from power-on/reset to the next power-off/reset). When the limit of 8 is reached and a disk failure occurs, the standby replacement will occur but will not be recorded in the replacement table.

To clear the "standby replacement" table, reboot the system from a DOS bootable floppy, run the configuration utility and select the option 'view/update configuration' from the main menu. A red box labeled 'Drive Remap List' will be displayed. Selecting the box will allow you to continue. You should save the configuration without making any changes, and exit the configuration utility. This will clear the replacement table. You may now proceed to boot your system and continue normal operations.

In normal use, the replacement table limit of 8 should not cause any problems. Assuming that a disk fails about once a year (drives we support generally come with a 5-year warranty), the system would run continuously for a minimum of 8 years before the table would need to be cleared.

Storage Device

A collective term for disks, tape transports, and other mechanisms capable of non-volatile data storage.

Stripe Order

The order in which SCSI disk drives appear within a drive group. This order must be maintained, and is critical to the controller's ability to "rebuild" failed drives.

Stripe Size

The size, in kilobytes (1024 bytes) of a single I/O operation. A stripe of data (data residing in actual physical disk sectors, which are logically ordered first to last) is divided over all disks in the drive group.

Stripe Width

The number of striped SCSI drives within a drive group.

Striping

The storing of a sequential block of incoming data across multiple SCSI drives in a group. For example, if there are 3 SCSI drives in a group, the data will be separated into blocks. Block 1 of the data will be stored on SCSI drive 1, block 2 on SCSI drive 2, block 3 on SCSI drive 3, block 4 on SCSI drive 1, block 5 on SCSI drive 2, and so on. This storage method increases the disk system throughput by ensuring a balanced load among all drives.

Sub-System Storage

A collection of disks providing data storage space to a system user.

-- T --

Terminator

A part used to end a SCSI bus.

Termination

A method of matching transmission impedance of a bus to eliminate signal reflections from the physical ends of the bus.

Throughput

The number of I/O requests satisfied per unit of time (usually per second).

TPC-C, Tpm-C

The Transaction Processing Performance Council (TPC) is a standards organization that measures transaction throughput of systems. One of their benchmarks is Tpm-C, which reflects price and performance metrics. TPC-C reflects new order transaction rate, a benchmark for transaction speed. Mylex products have won consistently high TPC-C results.

Transfer Rate

The rate at which data moves between the host computer and storage, input, or output devices, usually expressed as a number of characters per second.

-- U --

-- V --

-- W --

Write-Back Cache

A caching strategy whereby write operations result in a completion signal being sent to the host operating system as soon as the cache (not the disk drive) receives the data to be written. The target disk drive will receive the data at a more appropriate time in order to increase controller performance. An optional cache battery backup can be used to protect against data loss as a result of a power failure or system crash.

Write-Through Cache

A caching strategy whereby data is written to the SCSI drive before a completion status is returned to the host operating system. This caching strategy is considered more secure, since a power failure will be less likely to cause loss of data. However, a write through cache results in a slightly lower performance.

-- X --

XOR

Exclusive "Or", a computer language function that generates parity in RAID systems; "this or that but not both."

-- Y --

-- Z --

Darryl Peddle
Internet Marketing Specialist
CBL Rec. de Dados
Toll Free: 1.800.551.3917
www.cbltech.com