Intel SRCU31, Руководство пользователя

Страница: 28 / 182

28 Intel RAID SRCU31 Users Guide
Install a SCSI terminator at the end of the SCSI bus or terminate the last device on the bus by doing
one of the following:
•
Change a switch setting on the device’s switch block (typically on external storage cabinets).
•
Insert or remove the terminator resistors from the terminator sockets (typically on hard disks).
•
Insert a jumper on the device to enable automatic termination.
•
Install a terminator cap (for external SCSI devices).
•
Install a physical terminator onto the end of the SCSI connector.
Choosing a Configuration
How Many Hard Disk Drives Should Be Integrated Into the Disk Array?
The number of physical drives the Intel RAID Controller SRCU31 can run determines the
maximum number of physical drives in a disk array. The minimum number of hard disk drives
required for any array depends of the RAID level you wish to realize. The desired usable disk
space of the disk array as well as the issues discussed in the following sections (What Level of
Redundancy is Needed and Are Hot Fix Drives Needed) have a direct impact on the number of
physical hard disk drives needed for an array.
Table 1. RAID Level, Array Type, and Hard Disk Drive Requirements
RAID Level
Disk Array Type Minimum Number of
Hard Disk Drives Required
RAID 0 Data Striping 2
RAID 1 Disk Mirroring 2
RAID 4 Data Striping with Parity Drive 3
RAID 5 Data Striping with Striped Parity 3
RAID 10 Data Striping and Mirroring 4
What Level of Redundancy is Needed?
RAID 0 (data striping) provides no redundancy but has high data throughput. With RAID 1
(disk mirroring) the data is 100% redundant because it is mirrored on other SCSI hard disk drives.
RAID 1 provides the highest level of redundancy, but is also the most expensive solution. A
combination of RAID levels 0 and 1 is RAID 10. Two RAID 0 stripe sets are mirrored. If one
drive fails, the data is available on the mirrored drive.
With RAID 4 (data striping with a dedicated parity drive) and RAID 5 (data striping with parity
striped across all RAID disks), parity information is calculated from the data with a simple
mathematical operation (eXclusive OR, XOR), and stored either on one dedicated drive (RAID 4)
or to all drives (RAID 5). Should one drive fail, the data of the defective drive can be reconstructed
using the remaining user data and the previously calculated parity data. RAID levels 4, 5 and 10
can tolerate the failure of one drive just as RAID 1, however RAID 4, RAID 5, or RAID 10 are less
expensive because of their efficient ratio of available to installed capacity.