gpt4 book ai didi

java - ConcurrentHashMap 与 AtomicReference>

转载 作者:行者123 更新时间:2023-12-01 17:49:31 27 4
gpt4 key购买 nike

选择什么更好?为什么?该 map 用作临时存储。它将项目保留一段时间,然后刷新到数据库。

这是我使用原子引用实现的虚拟代码:

public class Service {
private final AtomicReference<Map<String, Entity>> storedEntities = new AtomicReference<>(new HashMap<>());
private final AtomicReference<Map<String, Entity>> newEntities = new AtomicReference<>(new HashMap<>());

private final Dao dao;

public Service(Dao dao) {
this.dao = dao;
}

@Transactional
@Async
public CompletableFuture<Void> save() {
Map<String, Entity> map = newEntities.getAndSet(new HashMap<>());
return dao.saveAsync(map.values());
}

@Transactional(readOnly = true)
@Async
public CompletableFuture<Map<String, Entity>> readAll() {
return dao.getAllAsync().thenApply(map -> {
storedEntities.set(map);
return map;
});
}

@Scheduled(cron = "${cron}")
public void refreshNow() {
save();
readAll();
}

public void addNewentity(Entity entity) {
newEntities.getAndUpdate(map -> {
map.put(entity.getHash(), entity);
return map;
});
}

public AtomicReference<List<Entity>> getStoredEntities() {
return storedEntities.get().values();
}

public AtomicReference<List<Entity>> getNewEntities() {
return newEntities.get().values();
}
}

正如我所说,我只需要保留数据一段时间,然后通过 cron 将其刷新到数据库。我对什么是更好的方法感兴趣 - AR 与 CHM?

最佳答案

我首先假设您需要跨多个线程共享访问某些资源(临时存储映射)。

简短回答:

使用ConcurrentHashMap。

长(呃)答案:

如果您的期望是通过使用 AtomicReference,您将收到一致或线程安全的 map View ,或者您的操作将以原子方式执行,那么您很可能是不正确的 - 但是,因为您没有提供您的用法示例,我不能完全肯定地说。

虽然性能不容忽视,但它不应该是您关心的主要问题 - 相反,您应该确保程序能够正确且按预期执行,然后寻求改进其性能。

如果您有多个线程读取和/或写入临时存储映射,则映射的内部状态保持一致和正确非常重要;您可以通过以atomic的方式实现您的操作来实现这一点。和线程安全,从这个意义上说,两者都是 ConcurrentHashMap或由 SynchronizedMap 包裹的 map 将实现这一目标。然而,上述每种方法都以不同的粒度实现了确保这一点的方法,并且由于 ConcurrentHashMap 采取了专门的(乐观)方法,与包装映射的朴素(悲观)方法相反,ConcurrentHashMap 较少容易受到资源争用的影响,并且可能是两者中性能更高的一个。

继续前进

我提供了下面这篇文章中讨论的三种机制的实现;除了 Java API 文档之外,还可以探索下面的代码。

import java.io.PrintStream;
import java.text.DecimalFormat;
import java.util.Collections;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import java.util.concurrent.ConcurrentHashMap;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.TimeUnit;
import java.util.concurrent.atomic.AtomicReference;
import java.util.function.BiFunction;
import java.util.stream.Collectors;
import java.util.stream.IntStream;

public class ConcurrencyExample {

interface TemporaryStorage<K, V> {

V compute(K key, BiFunction<? super K, ? super V, ? extends V> remapper);

V put(K key, V value);

V get(K key);

void clear();

@FunctionalInterface
interface UnitTest<K, V> {

void test(TemporaryStorage<K, V> store, K key);

}

}

static class ConcurrentHashMapTS<K, V> implements TemporaryStorage<K, V> {

private final Map<K, V> map = new ConcurrentHashMap<>();

@Override
public V compute(K key, BiFunction<? super K, ? super V, ? extends V> remapper) {
return map.compute(key, remapper);
}

@Override
public V put(K key, V value) {
return map.put(key, value);
}

@Override
public V get(K key) {
return map.get(key);
}

@Override
public void clear() {
map.clear();
}
}

static class AtomicReferenceHashMapTS<K, V> implements TemporaryStorage<K, V> {

private final AtomicReference<Map<K, V>> map = new AtomicReference<>(new HashMap<>());

@Override
public V compute(K key, BiFunction<? super K, ? super V, ? extends V> remapper) {
return map.get().compute(key, remapper);
}

@Override
public V put(K key, V value) {
return map.get().put(key, value);
}

@Override
public V get(K key) {
return map.get().get(key);
}

@Override
public void clear() {
map.get().clear();
}
}

static class MonitorLockedHashMapTS<K, V> implements TemporaryStorage<K, V> {

private final Map<K, V> map = new HashMap<>();
private final Object mutex = new Object(); // could use the map as the mutex

@Override
public V compute(K key, BiFunction<? super K, ? super V, ? extends V> remapper) {
synchronized (mutex) {
return map.compute(key, remapper);
}
}

@Override
public V put(K key, V value) {
synchronized (mutex) {
return map.put(key, value);
}
}

@Override
public V get(K key) {
synchronized (mutex) {
return map.get(key);
}
}

@Override
public void clear() {
synchronized (mutex) {
map.clear();
}
}
}

static class WrappedHashMapTS<K, V> implements TemporaryStorage<K, V> {

private final Map<K, V> map = Collections.synchronizedMap(new HashMap<>());

@Override
public V compute(K key, BiFunction<? super K, ? super V, ? extends V> remapper) {
return map.compute(key, remapper);
}

@Override
public V put(K key, V value) {
return map.put(key, value);
}

@Override
public V get(K key) {
return map.get(key);
}

@Override
public void clear() {
map.clear();
}
}

static class AtomicUnitTest implements TemporaryStorage.UnitTest<Integer, Integer> {

@Override
public void test(TemporaryStorage<Integer, Integer> store, Integer key) {
store.compute(key, (k, v) -> (v == null ? 0 : v) + 1);
}
}

static class UnsafeUnitTest implements TemporaryStorage.UnitTest<Integer, Integer> {

@Override
public void test(TemporaryStorage<Integer, Integer> store, Integer key) {
Integer value = store.get(key);
store.put(key, (value == null ? 0 : value) + 1);
}
}

public static class TestRunner {

public static void main(String... args) throws InterruptedException {
final int iterations = 1_000;
final List<Integer> keys = IntStream.rangeClosed(1, iterations).boxed().collect(Collectors.toList());

final int expected = iterations;

for (int batch = 1; batch <= 5; batch++) {

System.out.println(String.format("--- START BATCH %d ---", batch));

test(System.out, new ConcurrentHashMapTS<>(), new AtomicUnitTest(), keys, expected, iterations);
test(System.out, new ConcurrentHashMapTS<>(), new UnsafeUnitTest(), keys, expected, iterations);

test(System.out, new AtomicReferenceHashMapTS<>(), new AtomicUnitTest(), keys, expected, iterations);
test(System.out, new AtomicReferenceHashMapTS<>(), new UnsafeUnitTest(), keys, expected, iterations);

test(System.out, new MonitorLockedHashMapTS<>(), new AtomicUnitTest(), keys, expected, iterations);
test(System.out, new MonitorLockedHashMapTS<>(), new UnsafeUnitTest(), keys, expected, iterations);

test(System.out, new WrappedHashMapTS<>(), new AtomicUnitTest(), keys, expected, iterations);
test(System.out, new WrappedHashMapTS<>(), new UnsafeUnitTest(), keys, expected, iterations);

System.out.println(String.format("--- END BATCH %d ---", batch));

System.out.println();

}
}

private static <K, V> void test(PrintStream printer, TemporaryStorage<K, V> store, TemporaryStorage.UnitTest<K, V> work, List<K> keys, V expected, int iterations) throws InterruptedException {
test(printer, store, work, keys, expected, iterations, Runtime.getRuntime().availableProcessors() * 4);
}

private static <K, V> void test(PrintStream printer, TemporaryStorage<K, V> store, TemporaryStorage.UnitTest<K, V> work, List<K> keys, V expected, int iterations, int parallelism) throws InterruptedException {
final ExecutorService workers = Executors.newFixedThreadPool(parallelism);
final long start = System.currentTimeMillis();

for (K key : keys) {
for (int iteration = 1; iteration <= iterations; iteration++) {

workers.execute(() -> {
try {
work.test(store, key);
} catch (Exception e) {
//e.printStackTrace(); //thrown by the AtomicReference<Map<K, V>> implementation
}
});

}
}

workers.shutdown();
workers.awaitTermination(Long.MAX_VALUE, TimeUnit.DAYS);

final long finish = System.currentTimeMillis();

final DecimalFormat formatter = new DecimalFormat("###,###");

final long correct = keys.stream().filter(key -> expected.equals(store.get(key))).count();

printer.println(String.format("Store '%s' performed %s iterations of %s across %s threads in %sms. Accuracy: %d / %d (%4.2f percent)", store.getClass().getSimpleName(), formatter.format(iterations), work.getClass().getSimpleName(), formatter.format(parallelism), formatter.format(finish - start), correct, keys.size(), ((double) correct / keys.size()) * 100));
}

}

}

关于java - ConcurrentHashMap<String, Ref> 与 AtomicReference<Map<String, Ref>>,我们在Stack Overflow上找到一个类似的问题: https://stackoverflow.com/questions/51951124/

27 4 0
Copyright 2021 - 2024 cfsdn All Rights Reserved 蜀ICP备2022000587号
广告合作:1813099741@qq.com 6ren.com