作业一：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？
随着并发压力的增加，系统的响应时间从平稳到逐步上升，当超过负载点后会激增，并最终导致系统崩溃；吞吐量起初会随着并发数增大，然后趋于平稳，当超过负载点后，吞吐量会下降，并最终导致系统崩溃。
这是由于起初在系统没有到达满负载的阶段，系统可以处于一个比较好和稳定的状态，当负责过大后，系统任意指标满负荷运作后，会导致大量请求无法及时处理，大量的堆积会耗尽资源，最终导致系统崩溃。
作业二：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。
代码如下：
PageViewRequestTask 负责执行 url 请求发送，并记录每个请求的响应时间。count 表示需要发送多少次请求。
public class PageViewRequestTask implements Runnable {
private String url; private int count; private CountDownLatch latch; List<Long> responseTimeList = new ArrayList<>();
@Override public void run() {
// send request one by one in one task for (int i = 0; i < count; i++) { try { System.out.println("task [" + Thread.currentThread().getId() + "] is sending request[" + i + "]"); long responseTime = sendRequest(url);
// add response time to result responseTimeList.add(responseTime); } catch (IOException e) { e.printStackTrace(); } }
// count down when this task is finished latch.countDown(); System.out.println("task [" + Thread.currentThread().getId() + "] is finished"); }
private long sendRequest(String url) throws IOException { HttpClient httpClient = HttpClients.createDefault(); HttpGet httpGet = new HttpGet(url);
long startTime = new Date().getTime();
// ignore error for this demo HttpResponse response = httpClient.execute(httpGet); //System.out.println("response status code: " + response.getStatusLine().getStatusCode());
return new Date().getTime() - startTime; }
public PageViewRequestTask(String url, int count, CountDownLatch latch) { this.url = url; this.count = count; this.latch = latch; }
public List<Long> getResponseTimeList() { return responseTimeList; }}
复制代码
StressTestLauncher 负责运行测试，它根据并发数创建线程，每个线程平均运行总请求数。最后打印 95%响应时间。
public class StressTestLauncher {
private static final String TEST_REQUEST_URL = "https://www.baidu.com"; private static final int TEST_REQUEST_COUNT = 100; private static final int TEST_REQUEST_CONCURRENCY_TASK_COUNT = 10; private static final int TEST_REQUEST_COUNT_PER_TASK = TEST_REQUEST_COUNT / TEST_REQUEST_CONCURRENCY_TASK_COUNT;
public static void main(String[] args) throws InterruptedException {
CountDownLatch latch = new CountDownLatch(TEST_REQUEST_CONCURRENCY_TASK_COUNT); List<PageViewRequestTask> tasks = new ArrayList<>(); List<Thread> taskThreads = new ArrayList<>();
// create threads according to concurrency count for (int i = 0; i < TEST_REQUEST_CONCURRENCY_TASK_COUNT; i++) { PageViewRequestTask task = new PageViewRequestTask(TEST_REQUEST_URL, TEST_REQUEST_COUNT_PER_TASK, latch); tasks.add(task); taskThreads.add(new Thread(task)); }
// start running stress tests for (Thread taskThread : taskThreads) { taskThread.start(); }
// wait until all threads finish latch.await(); System.out.println("Stress test is finished");
// merge all response results List<Long> responseTimes = new ArrayList<>(); for (PageViewRequestTask task : tasks) { responseTimes.addAll(task.getResponseTimeList()); }
// sort response time responseTimes.sort(Long::compareTo);
// calculate index of 95% response time, and print out the result int resultIndex = (int) Math.round(TEST_REQUEST_COUNT * 0.95) - 1; System.out.println("95% response time is: " + responseTimes.get(resultIndex) + " ms"); }}
复制代码
对以下网站进行了测试，每个网站发出 100 个请求，10 个并发:
95% response time is: 528 ms
95% response time is: 675 ms
95% response time is: 268 ms
评论