問57
頻出度★★★
デュアルシステムの説明はどれか。
平成24年度 秋期 ITパスポート試験 問57
テクノロジ系システム構成要素 › システムの評価指標
This question's explanation and your answers so far.
解説
問 57
Incorrectア
This describes a duplex system. A duplex system consists of two systems: a primary (active) system that handles normal operations, and a secondary (standby) system held in reserve in case the primary fails; when a fault occurs in the primary, operation is switched to the secondary, which takes over the processing. While on standby, the secondary is either kept running in the same state so that it can take over at any moment (hot standby) or made to do other work such as batch processing. The key point is that the two systems are not performing the same processing at the same time — one of them is on standby — so there is no function that compares the results to verify their correctness. It is therefore a different thing from a dual system, whose essence is cross-checking.
Incorrectイ
This describes a peer-to-peer (P2P) distributed processing system. It refers to a form in which the computers connected by a network stand in an equal relationship, sometimes acting as the requesting side and sometimes as the responding side, sharing resources and services with each other. This is a matter of how processing is distributed, not a duplication method for raising reliability. It contains none of the element essential to a dual system — performing the same processing twice and collating the results — so it is wrong.
Incorrectウ
This describes a hierarchical (superior-subordinate) distributed processing system, typified by the client/server system. It refers to a form in which roles are given a clear hierarchical relationship — a server that provides services and a client that uses them — and the processing is divided between them. As with イ, this too is a classification of distributed processing structure, not a method of duplicating a system to ensure reliability. It refers neither to cross-checking of results nor to switchover at the time of a fault, so it does not describe a dual system.
- エ
正解 The correct answer is エ. A dual system prepares two computer systems of identical configuration, has both of them carry out exactly the same processing at the same time, and compares (cross-checks) the two processing results against each other. If the results of the two systems agree, the result is accepted as correct and processing moves on to the next step. If the results differ, it can be judged that a fault or a malfunction has occurred somewhere; the abnormal system is disconnected, and processing can continue with the remaining normal system alone. This mechanism has two benefits. One is that it does not merely detect a failure but also confirms whether the answer produced is correct, so the reliability of the results themselves is high. The other is that processing does not stop even if one side fails, so availability (operating rate) is also high. On the other hand, because both systems are run at full capacity at all times, equipment costs and operating costs both become large. For that reason it is adopted for uses that demand extremely high reliability, where an error is directly linked to human life or to society, such as a bank online account system, a seat reservation system, or train operation control. The statement in the question, “two systems perform exactly the same processing and the results are cross-checked,” matches this definition exactly, no more and no less, so エ is correct.
ここがポイント
- Distinguishing a dual system from a duplex system — a dual system runs the same processing on two systems at the same time and cross-checks the results to guarantee reliability, whereas a duplex system prepares a primary (active) system and a secondary (standby) system and switches over when a fault occurs. Both are forms of duplication, but whether there is constant parallel processing and collation of results is the decisive difference.
- Duplication (reliability design) and distributed processing (system structure) are different viewpoints — peer-to-peer means an equal relationship and client/server means a clear hierarchical relationship, and these are classifications of the form of distributed processing, so the viewpoint being asked about differs from that of duplication methods that raise fault tolerance. In a terminology question, working out which viewpoint a term is a classification word for is the shortcut to the correct answer.
用語