架构师训练营 - 第七周 - 作业

狂奔嘀兔纸

关注

发布于: 2020 年 07 月 22 日

一.性能压测的时候，随着并发压力的增加，系统响应时间和吞吐量如何变化，为什么？

1.性能测试阶段：系统并发访问数在系统的指定目标之内，系统硬件资源充足，此时系统的响应时间短，系统吞吐量未到达系统上线，所以吞吐量处于线性增长趋势，响应时间比较短，增长缓慢。

2.负载测试阶段：并发用户数递增，系统硬件资源利用率激增，到硬件资源到达一定临界点时，等待资源分配，并发请求不能及时处理，如果继续增加并发数，系统吞吐量增加缓慢，达到系统临界点后系统吞吐量开始降低，响应时间快速增长。

3.压力测试阶段：系统资源使用率到达饱和状态之后，继续增加并发量，此时系统吞吐量快速降低，直到系统压跨，响应时间延长，系统不再处理任何请求。

二.用你熟悉的编程语言写一个 web 性能压测工具，输入参数：URL，请求总次数，并发数。输出参数：平均响应时间，95% 响应时间。用这个测试工具以 10 并发、100 次请求压测 www.baidu.com。

package com.zztest;
import java.util.ArrayList;import java.util.Collections;import java.util.List;import java.util.concurrent.CountDownLatch;import java.util.concurrent.ExecutorService;import java.util.concurrent.Executors;import java.util.concurrent.Semaphore;
/** * @author * @Title:null.java * @Package PACKAGE_NAME * @Description: * @date 2020/07/22 */public class test {    public static final String url = "https://www.baidu.com";    public static final int requestCount = 100;    public static final int concurrentThreadCount = 10;
    public static void main(String[] args) throws InterruptedException {
        List<Long> responseTimeList = Collections.synchronizedList(new ArrayList<>());        ExecutorService executorService = Executors.newCachedThreadPool();        CountDownLatch countDownLatch = new CountDownLatch(requestCount);        Semaphore semaphore = new Semaphore(concurrentThreadCount);        for (int i = 0; i < requestCount; i++) {            executorService.execute(() -> {                try {                    semaphore.acquire();                    long startTime = System.currentTimeMillis();
                    CommonUtil.get(url);
                    long endTime = System.currentTimeMillis();                    responseTimeList.add(endTime - startTime);
                    semaphore.release();                } catch (InterruptedException e) {                    e.printStackTrace();                } catch (Exception e) {                    e.printStackTrace();                }                countDownLatch.countDown();            });        }
        countDownLatch.await();        executorService.shutdown();

        BenchReport(responseTimeList);
    }
    private static void CountPercent95ReponseTime(List<Long> responseTimeList,double percent){        long percent95ResponseTime = 0;
        Collections.sort(responseTimeList);        double px = percent * (responseTimeList.size() - 1);        int i = (int) Math.floor(px);        double g = px - i;        if (g == 0) {            percent95ResponseTime = responseTimeList.get(0);        } else {            percent95ResponseTime = (long) ((1 - g) * responseTimeList.get(i) + g * responseTimeList.get(i + 1));        }        System.out.println("95%响应时间（毫秒）：" + percent95ResponseTime);    }
    private static void BenchReport(List<Long> timeList) {        long totalResponseTime = 0;        long averageResponseTime;

        for (Long time : timeList) {            totalResponseTime += time;        }
        averageResponseTime = Math.round(totalResponseTime / timeList.size());        System.out.println(url + "压测结束：");        System.out.println("总请求数：" + requestCount);        System.out.println("并发数：" + concurrentThreadCount);        System.out.println("平均响应时间（毫秒）：" + averageResponseTime);
        CountPercent95ReponseTime(timeList,0.95);    }}

复制代码

package com.zztest;
import okhttp3.*;import java.io.IOException;

/** * @author * @Title:null.java * @Package PACKAGE_NAME * @Description: * @date 2020/07/22 */
public class CommonUtil {    /**     * 给定特定的url，参数封装成一个map,get请求     * @param url     * @param params     * @return 响应字符串     */    private static OkHttpClient client = new OkHttpClient();
    public static String get(String url){        Request request = new Request.Builder()                .url(url)                .build();        try {            String res = client.newCall(request).execute().body().string();            return res;        } catch (IOException e) {            e.printStackTrace();        }        return null;    }}