Basic concepts about raid and SCSI

Raid Overview
The disk subsystem with embedded microprocessor is usually called the r a I D system. The total available capacity of the r a I D array is smaller than the total capacity of the member disk.
I. RAID 0 (Block) is a simple disk block without verification. In essence, it is not a real r a I d, because it does not provide any form of redundancy. If the disk with RAID 0 fails, data will be completely lost. To recover data in RAID 0, the only method is to use tape backup or image copy.
2. Raid 1 (image) is not a verified r a I D level.
Iii. Definition of raid 2 (parallel access to a proprietary disk) involves the error verification circuit in the r a I d controller. This feature has been integrated into the disk drive, although cheap, but not efficient. Therefore, raid 2 is not a product.
4. Parallel access to r a I d belongs to r a I D 3. The r a I D 3 subsystem stores data in blocks to all the drives in the array, and writes the verification data to another disk in the array, r a I d 3 is considered as a school primary a I D.
5. Raid 4 (using dedicated verification disk for independent access) is an independently accessed r a I D Implementation, which uses a dedicated verification disk. Unlike RAID 3, raid 4 has a larger number of blocks, enabling simultaneous processing of multiple I/O requests. Although it provides performance advantages for read requests, raid 4 has a high write overhead because the disk is accessed twice during each read, modification, and write cycle.
6. RAID 5 (independent access using distributed verification) is an independently accessed r a I D array, which verifies that data is distributed across all disks in the array. In other words, there is no dedicated verification disk, so there is no write Bottleneck like raid 4.
VII. Raid 6 (with dual-verified independent access) provides two levels of redundancy, that is, when two drives in the array fail, the array can continue to work.

 

Raid 1: we often call it a "disk image", which fully protects data by completely copying the same data on a hard disk in the array. If one hard disk is damaged, the other hard disk will provide accurate and identical data, and the RAID system will switch to the image hard disk for further use. For users, data is not lost.
The bad thing about this image system is that the data storage speed has not been improved and the disk utilization is low. However, it provides the simplest and most effective protection for managers. When a hard disk fails, the array management software will directly switch data requests to a valid hard disk.

RAID 3: RAID 3 distributes data across multiple drives in a staggered manner. A dedicated hard disk allows users to store parity data and recover and recreate error data.

RAID 5: RAID 5 is the most common configuration method. It is a data storage method with the parity data recovery function. In RAID 5, the parity data blocks are distributed in each hard disk in the array, so that the data connection is smoother.
If one of the hard disks is damaged, the parity data is used for Data Reconstruction. This is a common practice. The disadvantage of this method is that the read/write time of data is relatively long (two read/write operations must be performed when a group of data is written ). Its capacity is N-1 and the minimum must have three hard disks.
 

Disk Array Glossary
Array: Array

The disk array mode integrates the storage space of several disks to form a large, single, continuous storage space. The netraid controller uses its SCSI channel to combine multiple disks into a disk array. Simply put, an array is a disk system composed of multiple disks that work in parallel. Note that hot backup disks cannot be added to the array.

Array spanning: array spanning

An array overlay is a process of re-integrating the storage space in two, three, or four disk arrays to form a logical drive with a single continuous storage space. The netraid controller can span several consecutive arrays, but each array must consist of the same number of disks, and these arrays must have the same raid level. That is to say, the spanning array is a combination of several already formed arrays. Raid 1, RAID 3, and RAID 5 form raid 10 after they are crossed, raid 30 and raid 50.

Cache Policy: high-speed cache policy

The netraid controller has two high-speed cache policies: cached I/O (Cache I/O) and direct I/O (direct I/O ). Cache I/O always adopts read and write policies, and is often cached randomly during read operations. Direct I/O always reads new data directly from the disk. If a data unit is read repeatedly, a moderate read policy is selected, and the read data will be cached. Data enters the cache only when the read data is accessed repeatedly, but no data enters the cache in the completely random read state.

Capacity Expansion: Capacity Expansion

In Microsoft's Windows NT, 2000, or Novell's Netware 4.2 and 5 operating systems, you can increase the current volume capacity online. In Windows 2000 or NetWare 5, disable the virtual capacity option when preparing for online resizing. In Windows NT or Netware 4.2, the virtual capacity option must be available for online resizing.

In the quick Configuration tool of the netraid controller, when the virtual capacity option is set to available, the Controller creates a virtual disk space, and then the volume can expand the added physical disk to the virtual space by restructuring. Refactoring can only be performed on a single logical drive in a single array. You cannot use online resizing across arrays.

Channel: Channel

The electrical path that transmits data and control information between two disk controllers.

Format: Format

In the process of writing zero to all data areas of a physical drive (hard disk), formatting is a pure physical operation. At the same time, it checks the consistency of the hard disk media, and mark the unreadable and bad sectors. Most hard disks have been formatted at the factory, so they must be formatted only when an error occurs in the hard disk media.
Hot spare: Hot Standby

When a disk in use fails, a idle, powered, and standby disk will replace the faulty disk immediately. This method is hot backup. The hot backup disk does not store any user data. It can have up to 8 disks as hot backup disks. A hot backup disk can belong to a single Redundant Array or be part of the hot backup disk pool of the entire array. In a specific array, there can be only one hot backup disk.

When the disk fails, the Controller's firmware can automatically replace the faulty disk with a hot backup disk.AlgorithmRebuild the data originally stored on the faulty disk to the hot backup disk. Data can only be rebuilt from a redundant logical drive (except RAID 0), and the hot backup disk must have enough capacity. The system administrator can change the faulty disk and specify the new hot backup disk as the new one.

Hot Swap disk module: Hot Swap disk Mode

The Hot Swap mode allows the system administrator to change the faulty disk drive when the server is continuously powered and network services are not suspended. Since all power supplies and cable connections are integrated on the server's baseboard, Hot Swap mode can directly remove the disk from the drive cage slot, which is very easy to operate. Then insert the replaced Hot Swap disk into the slot. Hot Swap Technology works only when RAID 1, 3, 5, 10, 30, and 50 are configured.

I2o (Intelligent Input/Output): Intelligent Input and Output

Intelligent Input/output is an industrial standard. The architecture of the input/output subsystem is completely independent of the network operating system and does not require the support of external devices. I2o driverProgramIt can be divided into the operating system Services Module (osms) and hardware driver module (hdms ).

Initialization: initialization

Writes zero to the data area of the logical drive, and generates the corresponding parity to make the logical drive ready. Initialization will delete the previous data and generate parity. Therefore, the logical drive will perform consistency detection during this process. An array that has not been initialized cannot be used, because no parity zone is generated, the array will produce a consistency check error.

IOP (I/O processor): Input/Output Processor

The input and output processor is the instruction center of the netraid controller, which includes command processing, data transmission of PCI and SCSI bus, raid processing, and disk drive reconstruction, high-speed cache management, error recovery, and other functions.

Logical drive: logical drive

Virtual Drive in the array, which can occupy more than one physical disk. A logical drive splits an array or a disk that spans an array into contiguous buckets distributed across all disks in the array. The netraid controller can set up to eight logical drives of different capacities, and at least one logical drive must be set in each array. The input and output operations can only be performed when the logical drive is online.

Logical volume: logical volume

A Virtual Disk formed by a Logical Disk is also called a disk partition.

Mirroring ing: Image

A type of redundancy. An identical copy of data on one disk exists on another disk, that is, an image. Raid 1 and raid 10 use images.
Parity: parity bit

In data storage and transmission, an additional bit is added to the byte to verify errors. It usually generates a redundant data from two or more original data, and redundant data can be reconstructed from one original data. However, the parity data is not completely copied to the original data.

In raid, this method can be applied to all disk drives in the array. The parity bit can also form a dedicated parity method. In the dedicated parity, the parity data can be distributed across all the disks in the system. If a disk fails, you can use the data on other disks and the parity data to reconstruct the data on the faulty disk.

Power fail safeguard: power-down protection

When this item is set to available, all data will be stored on the disk during the rebuild process (not rebuild) until the rebuild is complete. In this way, if a power loss occurs during the reconstruction process, there will be no risk of data loss.

Raid: Independent redundant disk array

An independent redundant disk array is initially called a cheap redundant disk array (Redundant Array of Inexpensive Disks). It is an array consisting of multiple small capacity and independent hard disks, the integrated performance of arrays can exceed the performance of a single expensive large-capacity hard drive (sled. Because the raid disk sub-system operates on multiple disks in parallel, its input and output performance is improved compared with that of a single disk. The server regards the raid array as a single storage unit and accesses several disks at the same time, thus increasing the input/output rate.

RAID levels: Raid level

The raid level is the application of different redundancy types on the logical drive. It can improve the fault tolerance and performance of the logical drive, but also reduce the available capacity of the logical drive, each logical drive must specify a raid level.

Raid 1, 3 and 5 logical drives use a single array, and Appendix 1 describes their details. In short, RAID 0 is non-redundant and can be composed of one or more physical drives; RAID 1 is image redundancy, which requires two physical drives in an array; RAID 3 is dedicated parity redundancy, that is, all redundant data is stored on a dedicated disk. An array consists of at least three physical drives. RAID 5 is distributed parity redundancy, redundant data in the array is distributed across all disks in the array. One of its arrays requires at least three physical drives.

Raid 10, 30, and 50 are logical drives that span arrays. Table 2 describes scenarios that span disk arrays.

Read policy: Read Policy

The netraid Controller provides three read policies: Read-ahead (pre-read), normal (standard), and adaptive (moderate ).

During the pre-read operation, the Controller constantly reads unrequested data in advance, stores it in the memory, and expects the data to be used. Pre-reading can provide continuous data faster. When random data is accessed, the validity period is not good.

The standard policy does not use the pre-read method. This policy is most effective when most of the data read is random data.

The moderate policy is to use the pre-read method when the data on the last two disks accessed is stored on the continuous sector.

Ready state: Ready

The ready state is an available hard disk, which is neither online nor hot backup, and can be added to any array or specified as hot backup disk.
Rebuild: Rebuilding

In raid, 30, or 50 arrays, all data on a faulty disk is regenerated to the replaced disk. During disk reconstruction, the logical drive normally does not interrupt access requests to its data.

Rebuild rate: rebuilding Rate

Rebuild speed. Each controller is assigned a reconstruction rate, which reflects the percentage of IOP resources used in the reconstruction operation.

Reconstruct: refactoring

The process of reorganizing the data on the logical drive after changing the raid level.

SCSI disk status: SCSI disk status

SCSI disks (physical drives) can be in the following five States: Ready (ready), unconfigured power-on, and online (online; hot spare (hot backup). When a disk fails, the disk to be powered on will be used; failed (fault ), A disk error may cause a failure or the state of the drive offline using the netraid controller utility; rebuilding where the disk is recovering data from one or more key logical drives.

Stripe size: Strip capacity

The total amount of data written continuously on each disk, also known as "band depth ". You can specify the bandwidth of each logical drive from 2kb, 4kb, 8kb to 128kb. To achieve higher performance, you need to select the bandwidth equal to or less than the size of the operating system cluster. Large-capacity strips produce higher read performance, especially when reading continuous data. When reading random data, it is better to set the bandwidth to a smaller value. 8 MB memory is required if the KB strip is specified.

Striping: striping

Striping is a method of dividing continuous data into data blocks of the same size and writing each segment of data to different disks in the array. This technology is very useful. It provides much faster read/write speed than a single disk. when data is uploaded from the first disk, the second disk can determine the next data segment. Data striping is widely used in some modern databases and some raid hardware devices.

Virtual Sizing: virtual capacity

When this setting takes effect, for a logical drive, the Controller will report that the logical drive capacity is much larger than the actual physical capacity. "Virtual" space can be expanded online.

Write policy: write policy

When the processor writes data to the disk, the data is first written to the cache, and the processor may immediately read it again. Netraid has two write policies:

Write back: In the write-back state, data is written to the disk only when it is to be cleared from the cache. As the data read from the primary storage increases, you need to write data from the cache to the disk and write the updated data to the cache. Because a piece of data may be written to the cache for many times without disk access, the write-back efficiency is very high.

Write Through (full write). When data is input to the cache in the full write state, it is also written to the disk. Because the data has been copied to the disk, you can directly change the data to be replaced in the cache, so completely writing is much easier than writing back.

Storage Technology> disk image
Disk image is a simple form of device virtualization technology. Each I/O operation is performed on two disks,
The two disks look like a single disk. The disk image is also called r a I D 1.

Storage Technology> Storage Area Network (s a n)
The storage region Network (s a n) is a dedicated network that connects one or more systems to storage devices and subsystems,
S a N can be considered as a "backend" network responsible for storage and transmission, while the "front-end" Network (or Data Network) is responsible for normal
T c p/I p transmission.

Storage Technology> network connected storage (n a s)
Network Connection storage (n a s) indicates the network storage server product of the general control. A typical n a s is connected to a common network.
Generally, Ethernet is used to provide an integrated system with pre-configured disk capacity and storage management software.
Backup storage solution.

How to Set SCSI BIOS
The first time you enter the scsi bios settings screen, you will be confused by these complicated and understandable options; but don't be afraid. Next we will explain each setting in the scsi bios for you, it also describes its purpose and effect. Although the SCSI brands are numerous, the scsi bios settings are similar, here we refer to the adaptec AHA-2940U/uw scsi bios v1.32.
Run the "Configure/view host adaptec setting" command directly:
SCSI bus interface definitions
The functions of this project are mainly used to set the definitions of the SCSI interface card. There are three settings:
Host adaptec scsi id: Set the ID number of the SCSI interface card itself here (this card can be set to 0 ~ 15 ). The default value is 7. We strongly recommend that you do not change this default value.
SCSI parity checking: SCSI sets the "same-bit check" during data transmission. You can select enable or disabled ). If this option is enabled, you can confirm the correctness of data transmission. We strongly recommend that you enable this option unless the peripherals connected to SCSI do not support the same-bit check function.
Host adaptec SCSI termination: Set the terminal resistance of the SCSI interface card. You can select automatic, enable, or disable. If you only have an external device or an internal device, you must set it to enable. If both the internal and external devices have an internal device, you must set it to disable. If you change the SCSI internal and external Methods frequently, we recommend that you select automatic to automatically detect whether termination needs to be enabled or disabled.
Additional options
Although this item is called "APPEND", it is the core of the entire scsi bios settings. For example, set the Startup Device, synchronous and asynchronous, maximum transmission rate, scam's PNP support, CD-ROM boot, etc. So many settings may be a bit complicated, but you just need to follow our instructions, I believe I can understand it quite well.
Boot device options: Specifies which ID is used as the boot device. General settings can be applied to SCSI hard disk (including removable hard disk) or Mo, CD-ROM, etc., but after the settings still need to set the motherboard system BIOS boot sequence to SCSI priority. If you set the BIOS multiple Lun support to no in the SCSI device configuration, this change will not work. If your SCSI hard disk is ID-0 and the operating system is installed, you can use the above settings to Set ID-0 as the boot device.
SCSI Device Configuration: This menu is mainly used to set independent options for each SCSI peripheral device. You can set different parameters for peripherals of each ID. For example, set the maximum transmission rate of ID-5 to 40 MB/sec, and Id-0 to 20 mb/sec, the two peripherals can transmit data independently at different maximum transmission rates without instability of the SCSI bus. If you set the host adapter BIOS in advanced configuration options to disable, these options will not work.
Initiate sync negotiation: You can select YES or NO for synchronous transmission communication. If yes is selected, the data transmission between the SCSI interface card and the SCSI peripheral device will be performed in "synchronous" mode to accelerate the data transmission efficiency. If no is selected, disable this function. We recommend that you set this item to yes unless the peripheral device does not support this feature.
Maximum sync transfer rate: sets the maximum synchronization rate between the SCSI interface card and each peripheral. It is generally recommended that you set the maximum transmission rate to "maximum" here. Even if the peripherals do not support it, it will be automatically adjusted to the maximum rate that can be reached; if a peripheral cannot respond to the SCSI interface card with the maximum synchronous transmission rate, problems may occur during use. For example, if you use an interface card of the SCSI-3 specification and the peripherals belong to the SCSI-2 specification, it is generally recommended to manually adjust the maximum transfer rate of this peripheral to 10 MB/sec. The following are the maximum synchronization transfer rates that can be adjusted by the BIOS of various SCSI interface cards.
SCSI interface specification SCSI-2 ultra-SCSI ultra-wide SCSI Ultra2-Wide LVD/SE
Adjustable maximum synchronization rate: 10 MB/sec 20 mb/sec 40 MB/sec 80 Mb/sec
Enable disconnection: allows the SCSI interface card to temporarily interrupt the connection to the scsicard peripherals to release a wider SCSI channel to serve other SCSI peripherals. If you want to work on multiple SCSI peripherals in a "multi-task" way, using this function can improve the efficiency of multi-task operations. If you have more than two SCSI devices, you must set this item to yes. If you only have one SCSI peripheral device, you do not need to enable this function. Set it to No.
Initiate wide negotiaion: Set up broadband transmission communication. If set to yes, the SCSI peripherals and SCSI interface settings transmit data in a 16-bit wide channel. If set to no, the data is transmitted in 8-Bit mode. If your SCSI peripherals belong to the wide-SCSI series (68-pins, 16-bit), set this item to yes. Otherwise, 8-bit broadband transmission will be performed; if the peripheral device is SCSI-2, ultra SCSI (with a pin of 50-pins, 8-bit), we recommend that you set it to no, otherwise it may cause read problems.
Send start unit command: Set to send the startup command. Some SCSI peripherals need to use the SCSI interface card to send startup commands before the motor starts to run. Please refer to your SCSI peripherals manual. We recommend that you set this parameter to No. If your peripheral device does need to send a startup command to start the device, set this parameter to yes.
BiOS multiple Lun support: sets whether the SCSI interface card supports "Multi-logical unit numbers" (Lun, logical unit numbers ). If your peripheral device supports multiple Luns, you can set this item to yes. If such peripherals do not exist, set it to No. When a peripheral has multiple Luns, you can set the lun to be started in the boot device options boot Lun number project.
Include in BIOS scan: If this option is set to yes, the scsi bios scans the peripheral device at startup and controls it by the BIOS int13 extension. For example, if your SCSI peripherals are hard disks and this option is set to yes, the scsi bios will display "scsi bios installed successfully!" on the screen! ", And use the BIOS int13 Extension function of the SCSI interface card to control the SCSI hard disk without additional drivers. If it is set to no, the SCSI hard disk will still be detected at startup, however, the scsi bios will display "SCSI biosnot installed". At this time, the hard disk is not in the management scope of the extended BIOS int13 function. You naturally cannot see this SCSI hard disk in the operating system.
Advanced Configuration Options: Next we will introduce you to the advanced feature settings of the scsi bios. Please first return the scsi bios configuration screen, and then enter this option.
Reset SCSI bus at IC initialization: If this item is set to enabled, the signal in the SCSI channel will be reset every time the computer's power supply is re-enabled ), in this way, the SCSI interface card can be connected to all peripheral devices in the initial state. For example, when your SCSI CD-ROM is reading a problem and you cannot exit the disc, If you restart the computer, the blocked signal in the SCSI channel will be automatically cleared, restore to the initial connection status, and then the CD will exit smoothly!
Extended BIOS translation for DOS drives> 1 Gbyte: Set whether to support hard disks larger than 1 GB in the DOS environment. The default value is enable. Do not change the value.
Host adapter BIOS (configuration utility reserves BIOS space): If your SCSI peripheral hard disk needs to be started with a SCSI hard disk, set this option to enable; if SCSI peripherals only contain CD-Rom and scanners, you can set them to disable.
Support removable disks under BIOS as fixed disks: This option is especially for Mo or removable hard disks. If it is set to boot only, the boot disk (started with a mo or removable hard disk) is used as the hard disk; if it is set to all disks, all Mo or removable hard disks can be used as the hard disk; if it is set to disabled, Mo is restored to a removable storage device. Note that you must put the mo disk in a VM before starting the VM. Once you use the mo hard disk, if you quit the mo disk in the connected state, this will cause data loss.
Display message during BIOS initialization: If this option is set to enabled, the scsi bios will display press for SCSI select (TM) utility on the screen at startup !, Prompt you to follow this method to enter the scsi bios settings menu. If it is set to disable, this line of prompt will not be displayed, but you can start to use Ctrl + a key to Jin Nian-day microscsi BIOS menu.
BIOS support vfor bootable CD-ROM: If this option is set to enable, you can put the boot disc into the SCSI optical drive and start it with the SCSI Optical Drive.
BIOS support for int13 extensions: DOS or Windows 95 of the old version only supports the interruption of the traditional int13h Disk control, and supports a maximum of 1024 cylinder ). If you have installed the Windows 95 osr2 or Windows 98 operating system, you can support the new int13 extensions function. At this time, you can manage disks with a capacity of more than 8 GB (that is, disks with a capacity of> 1024cylinder ), however, you must set this item to enable, so that your operating system can manage hard disks larger than 8 GB.