微服务生态组件之Spring Cloud LoadBalancer详解和源码分析
Spring Cloud LoadBalancer
概述
Spring Cloud LoadBalancer目前Spring官方是放在spring-cloud-commons里，Spring Cloud最新版本为2021.0.2
Spring Cloud LoadBalancer 官网文档地址 Cloud Native Applications
Spring Cloud官网文档地址 Spring Cloud
一方面Netflix Ribbon停止更新，Spring Cloud LoadBalancer是Spring Cloud官方自己提供的客户端负载均衡器,抽象和实现，用来替代Ribbon。
- 常见负载均衡器分为服务端负载均衡器(如网关层均衡负载)和客户端层均衡负载。
- 网关层如硬件层面的F5或软件层面的LVS、或者nginx等。
- 客户端层就如Spring Cloud LoadBalancer，作为一个客户端去发现更新维护服务列表，自定义服务的均衡负载策略（随机、轮询、小流量的金丝雀等等）。
Spring Cloud提供了自己的客户端负载平衡器抽象和实现。对于负载均衡机制，增加了ReactiveLoadBalancer接口，并提供了基于round-robin轮询和Random随机的实现。为了从响应式ServiceInstanceListSupplier中选择实例，需要使用ServiceInstanceListSupplier。目前支持ServiceInstanceListSupplier的基于服务发现的实现，该实现使用类路径中的发现客户端从Service Discovery中检索可用的实例。
可以通过如下配置来禁用Spring Cloud LoadBalance
spring:
cloud:
loadbalancer:
enabled: false
入门示例
前面simple-ecommerce项目创建已在父Pom引入三大父依赖，详细可以看下前面的文章<
注意如果是Hoxton之前的版本，默认负载均衡器为Ribbon，需要移除Ribbon引用和增加配置spring.cloud.loadbalancer.ribbon.enabled: false。
如果是在Spring Boot项目中添加下面的启动器依赖，该starter也包含了Spring Boot Caching and Evictor.
org.springframework.cloud
spring-cloud-starter-loadbalancer
我们使用Spring官方提供了负载均衡的客户端之一RestTemplate，RestTemplate是Spring提供的用于访问Rest服务的客户端，RestTemplate提供了多种便捷访问远程Http服务的方法，能够大大提高客户端的编写效率。默认情况下，RestTemplate默认依赖jdk的HTTP连接工具。创建RestTemplateConfig配置类，标注 @LoadBalanced注解，默认使用的ReactiveLoadBalancer实现是RoundRobinLoadBalancer。
package cn.itxs.ecom.order.config;
import org.springframework.cloud.client.loadbalancer.LoadBalanced;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import org.springframework.web.client.RestTemplate;
@Configuration
public class RestTemplateConfig {
@LoadBalanced
@Bean
public RestTemplate restTemplate() {
return new RestTemplate() ;
}
}
订单微服务中订单控制器增加deductRest方法
package cn.itxs.ecom.order.controller;
import cn.itxs.ecom.commons.service.OrderService;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.web.bind.annotation.PathVariable;
import org.springframework.web.bind.annotation.RequestMapping;
import org.springframework.web.bind.annotation.RestController;
import org.springframework.web.client.RestTemplate;
/**
* @Name ：OrderController
* @Description ：订单控制器
* @Author ：itxs
* @Date ：2022/4/10 20:15
* @Version ：1.0
* @History ：
*/
@RestController
public class OrderController {
@Autowired
OrderService orderService;
@Autowired
private RestTemplate restTemplate;
@RequestMapping("/create/{userId}/{commodityCode}/{count}")
public String create(@PathVariable("userId") String userId,@PathVariable("commodityCode") String commodityCode, @PathVariable("count") int count){
return orderService.create(userId,commodityCode,count).toString();
}
@RequestMapping("/deductRest/{commodityCode}/{count}")
public String deductRest(@PathVariable("commodityCode") String commodityCode, @PathVariable("count") int count){
String url = "http://ecom-storage-service/deduct/"+commodityCode+"/"+count;
return restTemplate.getForObject(url, String.class);
}
}
前面server.port我们是放在Nacos配置中心里，这里我们注释Nacos配置中心的配置放在本地配置文件bootstrap.yml里，分别配置为4080、4081、4082启动3个库存服务实例，并启动订单微服务
server:
port: 4080
查看nacos服务管理-服务列表里服务详情，可以看到3个健康的库存实例和1个订单微服务实例
访问6次订单dedect接口：http://localhost:4070/deductRest/1001/1 ，从测试的结果也验证了LoadBalancer默认是轮询负载均衡策略。
负载均衡算法切换
创建自定义负载均衡配置类CustomLoadBalancerConfiguration
package cn.itxs.ecom.order.config;
import org.springframework.cloud.client.ServiceInstance;
import org.springframework.cloud.loadbalancer.core.RandomLoadBalancer;
import org.springframework.cloud.loadbalancer.core.ReactorLoadBalancer;
import org.springframework.cloud.loadbalancer.core.ServiceInstanceListSupplier;
import org.springframework.cloud.loadbalancer.support.LoadBalancerClientFactory;
import org.springframework.context.annotation.Bean;
import org.springframework.core.env.Environment;
public class CustomLoadBalancerConfiguration {
@Bean
ReactorLoadBalancer randomLoadBalancer(Environment environment,
LoadBalancerClientFactory loadBalancerClientFactory) {
String name = environment.getProperty(LoadBalancerClientFactory.PROPERTY_NAME);
return new RandomLoadBalancer(loadBalancerClientFactory
.getLazyProvider(name, ServiceInstanceListSupplier.class),
name);
}
}
RestTemplateConfig配置类LoadBalancerClient指定随机的配置类，value的值为提供者也即是库存微服务名称。
package cn.itxs.ecom.order.config;
import org.springframework.boot.web.client.RestTemplateBuilder;
import org.springframework.cloud.client.loadbalancer.LoadBalanced;
import org.springframework.cloud.loadbalancer.annotation.LoadBalancerClient;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import org.springframework.web.client.RestTemplate;
@Configuration
@LoadBalancerClient(value = "ecom-storage-service", configuration = CustomLoadBalancerConfiguration.class)
public class RestTemplateConfig {
@LoadBalanced
@Bean
public RestTemplate restTemplate(RestTemplateBuilder builder) {
return builder.build() ;
}
}
多次访问订单dedect接口测试确认已切换为随机负载均衡策略。
集成方式
官网提供3中集成Spring Cloud LoadBalancer的方式，除了第一种上面已使用过，还支持Spring Web Flux响应式编程，WebClient是从Spring WebFlux 5.0版本开始提供的一个非阻塞的基于响应式编程的进行Http请求的客户端工具。它的响应式编程的基于Reactor的。WebClient中提供了标准Http请求方式对应的get、post、put、delete等方法，可以用来发起相应的请求。
在订单微服务中引入spring-boot-starter-webflux依赖
org.springframework.boot
spring-boot-starter-webflux
订单微服务中增加WebClientConfig配置类
package cn.itxs.ecom.order.config;
import org.springframework.cloud.client.loadbalancer.LoadBalanced;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import org.springframework.web.reactive.function.client.WebClient;
@Configuration
public class WebClientConfig {
@LoadBalanced
@Bean
WebClient.Builder webClientBuilder() {
return WebClient.builder();
}
@Bean
WebClient webClient() {
return webClientBuilder().build();
}
}
订单微服务订单控制器中添加WebClient接口实现，代码如下
@Autowired
private WebClient webClient;
@RequestMapping(value = "/deductWebClient/{commodityCode}/{count}")
public Mono deductWebClient(@PathVariable("commodityCode") String commodityCode, @PathVariable("count") int count) {
String url = "http://ecom-storage-service/deduct/"+commodityCode+"/"+count;
// 基于WebClient
Mono result = webClient.get().uri(url)
.retrieve().bodyToMono(String.class);
return result;
}
重新启动订单微服务
访问订单控制器中的减库存WebClient接口，http://localhost:4070/deductWebClient/1001/1 ，结果返回成功
我们还可以配置基于过滤器的方式，通过WebClient使用ReactiveLoadBalancer。如果项目中添加了Spring Cloud LoadBalancer starter，并且Spring -webflux在类路径中，ReactorLoadBalancerExchangeFilterFunction则是自动配置的。
订单微服务订单控制器中添加WebClient使用ReactiveLoadBalancer接口实现，代码如下
@Autowired
private ReactorLoadBalancerExchangeFilterFunction lbFunction;
@RequestMapping(value = "/deductWebFluxReactor/{commodityCode}/{count}")
public Mono deductWebFluxReactor(@PathVariable("commodityCode") String commodityCode, @PathVariable("count") int count) {
String url = "/deduct/"+commodityCode+"/"+count;
Mono result = WebClient.builder().baseUrl("http://ecom-storage-service")
.filter(lbFunction)
.build()
.get()
.uri(url)
.retrieve()
.bodyToMono(String.class);
return result;
}
重新启动订单微服务
访问订单控制器中的减库存WebClient接口，http://localhost:4070/deductWebFluxReactor/1001/1 ，结果返回成功
关于LoadBalancer官网还提供很多其他功能，有兴趣可自行详细查阅和动手实验
原理
RestTemplate
Spring Cloud LoadBalancer源码分析我们先从RestTemplate负载均衡的简单实现来分析入手，除此之外其支持Spring Web Flux响应式编程的实现原理思想也是相同，都是通过客户端添加拦截器，在拦截器中实现负载均衡。从RestTemplate的源码中可以知道其继承自InterceptingHttpAccessor抽象类
而InterceptingHttpAccessor抽象类则提供了一个方法setInterceptors，用于设置拦截器，拦截器需要实现ClientHttpRequestInterceptor接口即可，在实际远程请求服务端接口之前会先调用拦截器的intercept方法。这里的拦截器相当于Servlet技术中的Filter功能
// 代码实现在抽象父类InterceptingHttpAccessor里
// RestTemplate.InterceptingHttpAccessor#setInterceptors
public void setInterceptors(List interceptors) {
Assert.noNullElements(interceptors, "'interceptors' must not contain null elements");
// Take getInterceptors() List as-is when passed in here
if (this.interceptors != interceptors) {
this.interceptors.clear();
this.interceptors.addAll(interceptors);
AnnotationAwareOrderComparator.sort(this.interceptors);
}
}
LoadBalancerAutoConfiguration
从官网可以知道Spring Cloud LoadBalancer放在spring-cloud-commons，因此也作为其核心的@LoadBalanced注解也就是由spring-cloud-commons来实现，依据SpringBoot自动装配的原理先查看依赖包的实现逻辑，不难发现spring-cloud-commons引入了自动配置类LoadBalancerAutoConfiguration和ReactorLoadBalancerClientAutoConfiguration。
当满足上述的条件时（@Conditional为条件注解），将自动创建LoadBalancerInterceptor并注入到RestTemplate中。
LoadBalancerLnterceptor
LoadBalancerInterceptor实现了ClientHttpRequestInterceptor接口，因此也实现intercept方法，用于实现负载均衡的拦截处理。
LoadBalancerClient
LoadBalancerClient用于进行负载均衡逻辑，继承自ServiceInstanceChooser接口，从服务列表中选择出一个服务地址进行调用。在LoadBalancerClient种存在两个execute()方法，均是用来执行请求的，reconstructURI()是用来重构URL。
对于LoadBalancerClient接口Spring Cloud LoadBalancer的提供默认实现为BlockingLoadBalancerClient
@SuppressWarnings({ "unchecked", "rawtypes" })
public class BlockingLoadBalancerClient implements LoadBalancerClient {
private final ReactiveLoadBalancer.Factory loadBalancerClientFactory;
/**
* @deprecated in favour of
* {@link BlockingLoadBalancerClient#BlockingLoadBalancerClient(ReactiveLoadBalancer.Factory)}
*/
@Deprecated
public BlockingLoadBalancerClient(LoadBalancerClientFactory loadBalancerClientFactory,
LoadBalancerProperties properties) {
this.loadBalancerClientFactory = loadBalancerClientFactory;
}
public BlockingLoadBalancerClient(ReactiveLoadBalancer.Factory loadBalancerClientFactory) {
this.loadBalancerClientFactory = loadBalancerClientFactory;
}
@Override
public T execute(String serviceId, LoadBalancerRequest request) throws IOException {
String hint = getHint(serviceId);
LoadBalancerRequestAdapter lbRequest = new LoadBalancerRequestAdapter<>(request,
new DefaultRequestContext(request, hint));
Set supportedLifecycleProcessors = getSupportedLifecycleProcessors(serviceId);
supportedLifecycleProcessors.forEach(lifecycle -> lifecycle.onStart(lbRequest));
ServiceInstance serviceInstance = choose(serviceId, lbRequest);
// 选择服务
if (serviceInstance == null) {
supportedLifecycleProcessors.forEach(lifecycle -> lifecycle.onComplete(
new CompletionContext<>(CompletionContext.Status.DISCARD, lbRequest, new EmptyResponse())));
throw new IllegalStateException("No instances available for " + serviceId);
}
return execute(serviceId, serviceInstance, lbRequest);
}
@Override
public T execute(String serviceId, ServiceInstance serviceInstance, LoadBalancerRequest request)
throws IOException {
DefaultResponse defaultResponse = new DefaultResponse(serviceInstance);
Set supportedLifecycleProcessors = getSupportedLifecycleProcessors(serviceId);
Request lbRequest = request instanceof Request ? (Request) request : new DefaultRequest<>();
supportedLifecycleProcessors
.forEach(lifecycle -> lifecycle.onStartRequest(lbRequest, new DefaultResponse(serviceInstance)));
try {
T response = request.apply(serviceInstance);
Object clientResponse = getClientResponse(response);
supportedLifecycleProcessors
.forEach(lifecycle -> lifecycle.onComplete(new CompletionContext<>(CompletionContext.Status.SUCCESS,
lbRequest, defaultResponse, clientResponse)));
return response;
}
catch (IOException iOException) {
supportedLifecycleProcessors.forEach(lifecycle -> lifecycle.onComplete(
new CompletionContext<>(CompletionContext.Status.FAILED, iOException, lbRequest, defaultResponse)));
throw iOException;
}
catch (Exception exception) {
supportedLifecycleProcessors.forEach(lifecycle -> lifecycle.onComplete(
new CompletionContext<>(CompletionContext.Status.FAILED, exception, lbRequest, defaultResponse)));
ReflectionUtils.rethrowRuntimeException(exception);
}
return null;
}
private Object getClientResponse(T response) {
ClientHttpResponse clientHttpResponse = null;
if (response instanceof ClientHttpResponse) {
clientHttpResponse = (ClientHttpResponse) response;
}
if (clientHttpResponse != null) {
try {
return new ResponseData(clientHttpResponse, null);
}
catch (IOException ignored) {
}
}
return response;
}
private Set getSupportedLifecycleProcessors(String serviceId) {
return LoadBalancerLifecycleValidator.getSupportedLifecycleProcessors(
loadBalancerClientFactory.getInstances(serviceId, LoadBalancerLifecycle.class),
DefaultRequestContext.class, Object.class, ServiceInstance.class);
}
@Override
public URI reconstructURI(ServiceInstance serviceInstance, URI original) {
return LoadBalancerUriTools.reconstructURI(serviceInstance, original);
}
@Override
public ServiceInstance choose(String serviceId) {
return choose(serviceId, REQUEST);
}
// 通过不同的负载均衡客户端实现选择不同的服务
@Override
public ServiceInstance choose(String serviceId, Request request) {
ReactiveLoadBalancer loadBalancer = loadBalancerClientFactory.getInstance(serviceId);
if (loadBalancer == null) {
return null;
}
Response loadBalancerResponse = Mono.from(loadBalancer.choose(request)).block();
if (loadBalancerResponse == null) {
return null;
}
return loadBalancerResponse.getServer();
}
private String getHint(String serviceId) {
LoadBalancerProperties properties = loadBalancerClientFactory.getProperties(serviceId);
String defaultHint = properties.getHint().getOrDefault("default", "default");
String hintPropertyValue = properties.getHint().get(serviceId);
return hintPropertyValue != null ? hintPropertyValue : defaultHint;
}
}
LoadBalancerClientFactory
BlockingLoadBalancerClient中持有LoadBalancerClientFactory通过调用其getInstance方法获取具体的负载均衡客户端。通过工厂类LoadBalancerClientFactory获取具体的负载均衡器实例，后面的loadBalancer.choose(request)调用其接口choose()方法实现根据负载均衡算法选择下一个服务器完成负载均衡，而ReactiveLoadBalancer getInstance(String serviceId) 有默认实现LoadBalancerClientFactory
LoadBalancerClientFactory客户端实现了不同的负载均衡算法，比如轮询、随机等。LoadBalancerClientFactory继承自NamedContextFactory，NamedContextFactory继承ApplicationContextAware，实现Spring ApplicationContext容器操作。
ReactiveLoadBalancer
ReactiveLoadBalancer负载均衡器实现服务选择，Spring Cloud Balancer中实现了轮询RoundRobinLoadBalancer、随机RandomLoadBalancer、NacosLoadBalancer算法。
LoadBalancerClientConfiguration
如果没有显式指定负载均衡算法，默认缺省值为RoundRobinLoadBalancer
@Bean
@ConditionalOnMissingBean
public ReactorLoadBalancer reactorServiceInstanceLoadBalancer(Environment environment,
LoadBalancerClientFactory loadBalancerClientFactory) {
String name = environment.getProperty(LoadBalancerClientFactory.PROPERTY_NAME);
return new RoundRobinLoadBalancer(
loadBalancerClientFactory.getLazyProvider(name, ServiceInstanceListSupplier.class), name);
}
LoadBalancerRequestFactory
LoadBalancerRequest工厂类调用createRequest方法用于创建LoadBalancerRequest。其内部持有LoadBalancerClient对象也即持有BlockingLoadBalancerClient。
在日常项目中，一般负载均衡都是结合Feign使用，后续我们有时间再来分析Feign整合LoadBalancer的自动配置类FeignLoadBalancerAutoConfiguration的实现
ReactorLoadBalancerClientAutoConfiguration
我们也抛一下基于WebClient的@Loadbalanced的流程的引入，首先声明负载均衡过滤器ReactorLoadBalancerClientAutoConfiguration是一个自动装配器类，在项目中引入了 WebClient 和 ReactiveLoadBalancer 类之后，自动装配流程就开始运行，它会初始化一个实现了 ExchangeFilterFunction 的实例，在后面该实例将作为过滤器被注入到WebClient。后续流程有兴趣再自行研究
自定义负载均衡器
从上面可以知道LoadBalancerClientFactory是创建客户机、负载均衡器和客户机配置实例的工厂。它根据客户端名称创建一个Spring ApplicationContext，并从中提取所需的bean。因此进入到LoadBalancerClientFactory类中，需要去实现它的子接口ReactorServiceInstanceLoadBalancer，因为去获取负载均衡器实例的时候，是通过去容器中查找ReactorServiceInstanceLoadBalancer类型的bean来实现的，可以参照RandomLoadBalancer实现代码
package org.springframework.cloud.loadbalancer.core;
import java.util.List;
import java.util.concurrent.ThreadLocalRandom;
import org.apache.commons.logging.Log;
import org.apache.commons.logging.LogFactory;
import reactor.core.publisher.Mono;
import org.springframework.beans.factory.ObjectProvider;
import org.springframework.cloud.client.ServiceInstance;
import org.springframework.cloud.client.loadbalancer.DefaultResponse;
import org.springframework.cloud.client.loadbalancer.EmptyResponse;
import org.springframework.cloud.client.loadbalancer.Request;
import org.springframework.cloud.client.loadbalancer.Response;
public class RandomLoadBalancer implements ReactorServiceInstanceLoadBalancer {
private static final Log log = LogFactory.getLog(RandomLoadBalancer.class);
private final String serviceId;
private ObjectProvider serviceInstanceListSupplierProvider;
/**
* @param serviceInstanceListSupplierProvider a provider of
* {@link ServiceInstanceListSupplier} that will be used to get available instances
* @param serviceId id of the service for which to choose an instance
*/
public RandomLoadBalancer(ObjectProvider serviceInstanceListSupplierProvider,
String serviceId) {
this.serviceId = serviceId;
this.serviceInstanceListSupplierProvider = serviceInstanceListSupplierProvider;
}
@SuppressWarnings("rawtypes")
@Override
public Mono> choose(Request request) {
ServiceInstanceListSupplier supplier = serviceInstanceListSupplierProvider
.getIfAvailable(NoopServiceInstanceListSupplier::new);
return supplier.get(request).next()
.map(serviceInstances -> processInstanceResponse(supplier, serviceInstances));
}
private Response processInstanceResponse(ServiceInstanceListSupplier supplier,
List serviceInstances) {
Response serviceInstanceResponse = getInstanceResponse(serviceInstances);
if (supplier instanceof SelectedInstanceCallback && serviceInstanceResponse.hasServer()) {
((SelectedInstanceCallback) supplier).selectedServiceInstance(serviceInstanceResponse.getServer());
}
return serviceInstanceResponse;
}
private Response getInstanceResponse(List instances) {
if (instances.isEmpty()) {
if (log.isWarnEnabled()) {
log.warn("No servers available for service: " + serviceId);
}
return new EmptyResponse();
}
int index = ThreadLocalRandom.current().nextInt(instances.size());
ServiceInstance instance = instances.get(index);
return new DefaultResponse(instance);
}
}
保留核心实现进行简单仿写如下
package cn.itxs.ecom.order.config;
import java.util.List;
import java.util.Random;
import org.springframework.beans.factory.ObjectProvider;
import org.springframework.cloud.client.ServiceInstance;
import org.springframework.cloud.loadbalancer.core.ReactorServiceInstanceLoadBalancer;
import org.springframework.cloud.loadbalancer.core.ServiceInstanceListSupplier;
import org.springframework.cloud.client.loadbalancer.DefaultResponse;
import org.springframework.cloud.client.loadbalancer.EmptyResponse;
import org.springframework.cloud.client.loadbalancer.Request;
import org.springframework.cloud.client.loadbalancer.Response;
import reactor.core.publisher.Mono;
public class ItxsRandomLoadBalancerClient implements ReactorServiceInstanceLoadBalancer {
// 服务列表
private ObjectProvider serviceInstanceListSupplierProvider;
public ItxsRandomLoadBalancerClient(ObjectProvider serviceInstanceListSupplierProvider) {
this.serviceInstanceListSupplierProvider = serviceInstanceListSupplierProvider;
}
@Override
public Mono> choose(Request request) {
ServiceInstanceListSupplier supplier = serviceInstanceListSupplierProvider.getIfAvailable();
return supplier.get().next().map(this::getInstanceResponse);
}
/**
* 使用随机数获取服务
* @param instances
* @return
*/
private Response getInstanceResponse(
List instances) {
System.out.println("ItxsRandomLoadBalancerClient start");
if (instances.isEmpty()) {
return new EmptyResponse();
}
System.out.println("ItxsRandomLoadBalancerClient random");
// 随机算法
int size = instances.size();
Random random = new Random();
ServiceInstance instance = instances.get(random.nextInt(size));
return new DefaultResponse(instance);
}
}
将上面CustomLoadBalancerConfiguration替换为如下内容
package cn.itxs.ecom.order.config;
import org.springframework.beans.factory.ObjectProvider;
import org.springframework.cloud.loadbalancer.core.ReactorServiceInstanceLoadBalancer;
import org.springframework.cloud.loadbalancer.core.ServiceInstanceListSupplier;
import org.springframework.context.annotation.Bean;
public class CustomLoadBalancerConfiguration {
@Bean
public ReactorServiceInstanceLoadBalancer customLoadBalancer(ObjectProvider serviceInstanceListSupplierProvider) {
return new ItxsRandomLoadBalancerClient(serviceInstanceListSupplierProvider);
}
}
启动库存微服务和订单微服务，访问http://localhost:4070/deductRest/1001/1 ，控制台已打印自定义ItxsRandomLoadBalancerClient中的日志和成功访问结果
如果本文对你有帮助，别忘记给我个3连 ，点赞，转发，评论，
咱们下期见