情報処理システムの処理方式を図のように分類したとき,水平負荷分散システムを説明したものはどれか。 - 情報処理システム - 集中処理システム - 分散処理システム - 水平分散システム - 水平機能分散システム - 水平負荷分散システム - 垂直分散システム - 垂直機能分散システム - 垂直負荷分散システム
平成26年度 秋期 ITパスポート試験 問48
This question's explanation and your answers so far.
問 48
Incorrectア
This describes a client/server system, which corresponds to a vertical functional distribution system in the figure. The PC (client) handles input and display while the server handles data processing and output, so computers on an upper and a lower level share different roles. Because there is a higher/lower relationship it is “vertical,” and because what is divided is the role it is “functional” distribution. This differs from horizontal load distribution, where peer computers share the same role.
Incorrectイ
This is a configuration in which dedicated computers such as a print server and a mail server are provided separately from the computer that runs business applications, and it corresponds to a horizontal functional distribution system in the figure. The servers are peers, so it is “horizontal,” but each one handles a different role such as printing or mail, so it is “functional” distribution. It does not run the same application on multiple computers to share the processing volume, so it is not load distribution.
Incorrectウ
This is a configuration in which each branch's computer processes the sales data within that branch and the head office's computer collects them and processes the overall sales data; it corresponds to a vertical load-distribution system in the figure. Because there is a clear higher/lower relationship (hierarchy) between the head office and the branches it is “vertical,” and because the same kind of work, processing sales data, is shared across the levels it is “load” distribution. It can be told apart from horizontal load distribution, which assumes a relationship among peers, by whether or not a hierarchy exists.
- エ
正解 The correct answer is エ. The classification in the figure is built on two axes. The first axis, “horizontal vs. vertical,” is whether the computers that handle the processing have a hierarchical (higher/lower) relationship: if they are peers it is horizontal distribution, and if there is a parent-child style hierarchy it is vertical distribution. The second axis, “functional vs. load,” indicates whether what is being divided up is the role (function) or the amount of work (load). A horizontal load-distribution system is therefore a method in which multiple peer computers take on the same role and share the volume of processing. Choice エ, “a method in which multiple computers run the same application so that processing does not concentrate on a single computer,” satisfies both conditions: the same application means peers with the same role (horizontal), and preventing processing from concentrating means sharing the amount of work (load distribution). So it fits the description of a horizontal load-distribution system exactly. A typical example is placing several Web servers side by side and using a load balancer (load-distribution device) to spread the accesses among them; even if one machine fails, the remaining ones can keep processing, so this also improves availability.
- Organize distributed processing systems along two axes. “Horizontal vs. vertical” asks whether the computers have a higher/lower relationship (hierarchy) — peers means horizontal, a hierarchy means vertical — and “functional vs. load” indicates whether what is divided is the role or the amount of work. Combining the two axes gives four types: horizontal functional, horizontal load, vertical functional, and vertical load.
- Horizontal load distribution is a method in which multiple peer computers run the same application and share the processing volume (for example, spreading accesses across several Web servers with a load balancer). It not only prevents processing from concentrating and thus improves performance, but also keeps processing going on the remaining machines if one stops, so it helps improve availability as well. It is useful to memorize it in contrast with client/server, which is vertical functional distribution, and with the pattern where a head office and its branches share the same kind of processing, which is vertical load distribution.