“Load balancing” can be explained as maintaining any workload. For example if you have to serve 1000 breakfast in the morning you can divide the work among 2-3 or more caterers to lower the delivery/preparation time.
In the computer world, same logic applies, if you want to deliver fast, you can divide the work, for example for a website we can have 5-10 webserver, this way website will be delivered faster(especially during high traffic), this is server side.
I want to talk about client side load balancing and how this works, for example you are querying backend for every request, and you have multiple servers doing the backend work, for this I will use WHOIS service
whois1.service.com whois2.service.com
We can load balance every request in these two servers, most used algorithm and default round robin meaning every request will go to one than another, like who1, who2 who1 and so on, this can be done on the client side with “Spring Cloud LoadBalancer”. we need to configure 3 classes
- ServiceInstanceListSupplier for defining the backend addresses and ports
- Bean ServiceInstanceListSupplier for supplying the list of end points
- @LoadBalancerClient which is the webclient will be used for backend calls
Now, we have the whole setup for the backend
public class RestCaller implements ServiceInstanceListSupplier { private final String serviceId; public RestCaller(String serviceId) { this.serviceId = serviceId; } @Override public String getServiceId() { return serviceId; } @Override public Flux<List<ServiceInstance>> get() { return Flux.just(Arrays.asList( new DefaultServiceInstance(serviceId, serviceId, "abc.cyclic.app", 443, true), new DefaultServiceInstance(serviceId, serviceId, "someting.koyeb.app", 443, true), new DefaultServiceInstance(serviceId, serviceId, "127.0.0.1", 8080, false) )); } } @Configuration public class RestCallerConfiguration { @Autowired @Qualifier("webClientBuilder") WebClient.Builder webClientBuilder; @Bean @Primary ServiceInstanceListSupplier serviceInstanceListSupplier(ConfigurableApplicationContext ctx) { return ServiceInstanceListSupplier .builder() .withRetryAwareness() .withHealthChecks(webClientBuilder.build()) .withBase(new RestCaller("whoisService")) .build(ctx); } } @Configuration @LoadBalancerClient(name = "whoisService", configuration = RestCallerConfiguration.class) public class WebClientConfig { @LoadBalanced @Bean @Qualifier("loadBalancedWebClientBuilder") WebClient.Builder loadBalancedWebClientBuilder() { return WebClient.builder(); } @Bean @Qualifier("webClientBuilder") WebClient.Builder webClientBuilder() { return WebClient.builder(); } }
This can now be used in any @Component or @Service.
@Component public class WhoisCaller { @Autowired @Qualifier("loadBalancedWebClientBuilder") WebClient.Builder webClientBuilder; public String getWhois(String ip, String source) { String url = "http://whoisService/whois?ip=" + ip + "&source=" + source; return webClientBuilder.build() .get() .uri(url) .retrieve() .bodyToMono(String.class) .timeout(Duration.ofMinutes(1)) .doOnError(e -> { WebClientResponseException e1 = (WebClientResponseException) e; log.error(e1.getRequest().getURI().getHost()); }) .block(); } }
References:
- My question in SO to solve how to implement client side load balancing with spring
- Spring Cloud LoadBalancer
The above is the detailed content of What is load balancing and how to do it on client side. For more information, please follow other related articles on the PHP Chinese website!

Hot AI Tools

Undress AI Tool
Undress images for free

Undresser.AI Undress
AI-powered app for creating realistic nude photos

AI Clothes Remover
Online AI tool for removing clothes from photos.

Clothoff.io
AI clothes remover

Video Face Swap
Swap faces in any video effortlessly with our completely free AI face swap tool!

Hot Article

Hot Tools

Notepad++7.3.1
Easy-to-use and free code editor

SublimeText3 Chinese version
Chinese version, very easy to use

Zend Studio 13.0.1
Powerful PHP integrated development environment

Dreamweaver CS6
Visual web development tools

SublimeText3 Mac version
God-level code editing software (SublimeText3)

Hot Topics

The difference between HashMap and Hashtable is mainly reflected in thread safety, null value support and performance. 1. In terms of thread safety, Hashtable is thread-safe, and its methods are mostly synchronous methods, while HashMap does not perform synchronization processing, which is not thread-safe; 2. In terms of null value support, HashMap allows one null key and multiple null values, while Hashtable does not allow null keys or values, otherwise a NullPointerException will be thrown; 3. In terms of performance, HashMap is more efficient because there is no synchronization mechanism, and Hashtable has a low locking performance for each operation. It is recommended to use ConcurrentHashMap instead.

Java uses wrapper classes because basic data types cannot directly participate in object-oriented operations, and object forms are often required in actual needs; 1. Collection classes can only store objects, such as Lists use automatic boxing to store numerical values; 2. Generics do not support basic types, and packaging classes must be used as type parameters; 3. Packaging classes can represent null values ??to distinguish unset or missing data; 4. Packaging classes provide practical methods such as string conversion to facilitate data parsing and processing, so in scenarios where these characteristics are needed, packaging classes are indispensable.

StaticmethodsininterfaceswereintroducedinJava8toallowutilityfunctionswithintheinterfaceitself.BeforeJava8,suchfunctionsrequiredseparatehelperclasses,leadingtodisorganizedcode.Now,staticmethodsprovidethreekeybenefits:1)theyenableutilitymethodsdirectly

The JIT compiler optimizes code through four methods: method inline, hot spot detection and compilation, type speculation and devirtualization, and redundant operation elimination. 1. Method inline reduces call overhead and inserts frequently called small methods directly into the call; 2. Hot spot detection and high-frequency code execution and centrally optimize it to save resources; 3. Type speculation collects runtime type information to achieve devirtualization calls, improving efficiency; 4. Redundant operations eliminate useless calculations and inspections based on operational data deletion, enhancing performance.

Instance initialization blocks are used in Java to run initialization logic when creating objects, which are executed before the constructor. It is suitable for scenarios where multiple constructors share initialization code, complex field initialization, or anonymous class initialization scenarios. Unlike static initialization blocks, it is executed every time it is instantiated, while static initialization blocks only run once when the class is loaded.

Factory mode is used to encapsulate object creation logic, making the code more flexible, easy to maintain, and loosely coupled. The core answer is: by centrally managing object creation logic, hiding implementation details, and supporting the creation of multiple related objects. The specific description is as follows: the factory mode handes object creation to a special factory class or method for processing, avoiding the use of newClass() directly; it is suitable for scenarios where multiple types of related objects are created, creation logic may change, and implementation details need to be hidden; for example, in the payment processor, Stripe, PayPal and other instances are created through factories; its implementation includes the object returned by the factory class based on input parameters, and all objects realize a common interface; common variants include simple factories, factory methods and abstract factories, which are suitable for different complexities.

InJava,thefinalkeywordpreventsavariable’svaluefrombeingchangedafterassignment,butitsbehaviordiffersforprimitivesandobjectreferences.Forprimitivevariables,finalmakesthevalueconstant,asinfinalintMAX_SPEED=100;wherereassignmentcausesanerror.Forobjectref

There are two types of conversion: implicit and explicit. 1. Implicit conversion occurs automatically, such as converting int to double; 2. Explicit conversion requires manual operation, such as using (int)myDouble. A case where type conversion is required includes processing user input, mathematical operations, or passing different types of values ??between functions. Issues that need to be noted are: turning floating-point numbers into integers will truncate the fractional part, turning large types into small types may lead to data loss, and some languages ??do not allow direct conversion of specific types. A proper understanding of language conversion rules helps avoid errors.
