問86
頻出度★★★
システムの性能を評価する指標と方法に関する次の記述中のa～cに入れる字句の適切な組合せはどれか。 利用者が処理依頼を行ってから結果の出力が終了するまでの時間を a タイム，単位時間当たりに処理される仕事の量を b という。また，システムの使用目的に合致した標準的なプログラムを実行してシステムの性能を評価する方法を c という。
平成22年度 秋期 ITパスポート試験 問86
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 86
Incorrectア
This is a doubly wrong option in which a and b are swapped and c is wrong as well. The word that goes into a must express “time,” but throughput is the amount processed per unit of time, not a time, and there is no metric called “throughput time” either. Putting turnaround into b, which describes “the amount of work processed per unit of time,” also goes against the definition. The simulation in c is a method that models the target system and runs it imitatively on a computer to predict its behavior or performance; it is not a method of measuring by running a standard program on a real machine.
Incorrectイ
The benchmark test in c is correct, but a and b are swapped, so this option is inappropriate. Putting “throughput” into a means applying a term for an “amount” — the amount processed per unit of time — to the explanation of the elapsed “time” from the request until the result output finishes, which does not match the definition. Putting “turnaround” into b is the same kind of error, applying a term for time to the explanation of an amount. This is the typical trap answer for this question: it catches anyone who chooses by looking only at c.
Incorrectウ
a = Turnaround and b = Throughput are correct as far as they go, but c is wrong. Simulation is an evaluation method that models a system and runs it imitatively on a computer to predict the performance or the queueing behavior of a configuration that does not yet exist. The c in this question clearly states that the evaluation is done by “running a standard program,” which refers to a method of running and measuring on the actual system, so what belongs here is a benchmark test. Precisely because a and b are right, this is the most confusing option.
- エ
正解 The correct answer is エ (a: Turnaround, b: Throughput, c: Benchmark test). Check each blank against the definitions one by one. (a) The time from when a user issues a processing request until all of the result output has finished is called turnaround time. It is a metric of “time” that covers the whole round trip: the input and transmission time of the request data, the waiting time in the processing queue, the processing time on the CPU, and the output and printing time of the result. (b) The amount of work processed per unit of time is called throughput. It is a metric of “amount” such as how many items per hour or how many jobs per second, and it expresses the size of the processing capacity of the whole system. (c) The method of actually running a standard program (a benchmark program) that matches the intended purpose of the system, and evaluating the performance of the system by measuring the time it takes or the number of items processed, is called a benchmark test. The options split into two groups by how a and b are arranged, so first narrow them down by whether a is a word that expresses “time.” Throughput is an amount, and there is no metric called “throughput time,” so what goes into a is turnaround, which leaves ウ and エ as candidates. Next, c is a method that actually runs a program and measures it, not a model-based simulation, so it is a benchmark test. These two steps of narrowing down determine the answer uniquely as エ.
ここがポイント
- Distinguish the two pillars of performance metrics by their definitions. Turnaround time is the total elapsed time from the processing request until the result output has finished (input, waiting, processing, and output are all included), and its unit is time. The similar term response time is the time from making the request until the first response begins to come back; it differs in that it does not include the completion of the output. Throughput is the amount of work that can be processed per unit of time (items per second, jobs per hour, and so on), and its unit is an amount. Sorting terms by whether they are a metric of time or a metric of amount is exactly the key to solving this question.
- Distinguish how the performance evaluation methods are used. A benchmark test runs a standard program (a benchmark program) that matches the intended purpose on the actual system, and compares and evaluates performance using measured values. Simulation models the system and runs it imitatively to predict performance; it is used at stages such as design, where a real machine does not yet exist. Also learn monitoring (a method of measuring the resource usage of a system that is in operation) along with these.
用語