問82
頻出度★★☆
RAID1 (ミラーリング) の特徴として, 適切なものはどれか。
平成23年度 秋期 ITパスポート試験 問82
テクノロジ系システム構成要素 › システムの構成
This question's explanation and your answers so far.
解説
問 82
- ア
正解 RAID1 is the scheme called mirroring: it writes “exactly the same data” to two or more hard disks so that the data is duplicated. Even if one of the disks fails, the other still holds identical content, so the system can keep running without being stopped, and after the failed disk is replaced the data can be copied again to restore the redundant configuration. This property of “tolerating a failure by keeping a copy of the same data” matches the statement that it improves data availability (being able to keep using the data), so ア is correct. Note also that, as the price of this redundancy, the usable effective capacity is only about half of the total installed disk capacity in a two-disk configuration (about 1/n for n-way mirroring), and the write operation itself does not become faster — an important characteristic of RAID1.
Incorrectイ
Connecting several hard disks and treating them as a single virtual drive with their combined capacity describes JBOD (spanning / linear concatenation), not RAID1. It keeps no copy of the data, so if one disk fails the data on that disk is lost, and availability is in fact lower than that of a single disk. Note that it merely combines capacity and is also a different thing from striping, where data is split and written in parallel as in RAID0.
Incorrectウ
Splitting one piece of data and writing it to several hard disks in parallel to speed up writing describes RAID0 (striping). The speed does go up, but because it keeps no copy of the same data it has no redundancy, and if even one disk fails the whole set of data is lost. Its purpose is the exact opposite of RAID1, which improves availability.
Incorrectエ
Writing split data together with error-correcting parity information distributed across three or more hard disks describes RAID5. Parity allows the data to be restored for a failure of up to one disk, so availability is high, but it is not mirroring that copies the whole of the same data, and it also differs from RAID1 in requiring at least three disks.
ここがポイント
- RAID1 (mirroring) is a redundancy scheme that writes the same data to several disks in duplicate; its aim is higher availability, not write speed. The price is that the effective capacity is about half of the total capacity in a two-disk configuration (about 1/n for n-way mirroring).
- Telling the main RAID levels apart: RAID0 = striping (split and written in parallel for speed, no redundancy), RAID1 = mirroring (a copy raises availability, at least two disks), RAID5 = distributed parity (at least three disks, balancing availability and performance). Simply adding capacities together by concatenation is JBOD / spanning, which is not RAID redundancy.
用語