Stackify is now BMC. Read theBlog

How Spring Web MVC Really Works

By: Eugen
  |  February 27, 2024
How Spring Web MVC Really Works


This is an in-depth look at the powerful features and internal workings of Spring Web MVC, which is a part of the Spring Framework.

The source code for this article is available over on GitHub.

Project Setup

Throughout this article, we’ll use the latest and greatest Spring Framework 5. We’re focusing here on the Spring’s classic web stack, which has been available from the very first versions of the framework and is still the primary way of building web applications with Spring. 

For starters, to set up your test project, you’ll use Spring Boot and some of its starter dependencies; you’ll also need to define the parent:



Note that, in order to use Spring 5, you need to also use Spring Boot 2.x. At the time of writing, this is a milestone release, available in the Spring Milestone Repository. Let’s add this repository to your Maven project:

        <name>Spring Milestones</name>

You can check out the current version of Spring Boot on Maven Central.

Sample Project

To understand how Spring Web MVC works, you’ll implement a simple application with a login page. To show the login page, create a @Controller-annotated class InternalController with a GET mapping for the context root.

The hello() method is parameterless. It returns a String which is interpreted by Spring MVC as a view name (in our case, the login.html template):

import org.springframework.web.bind.annotation.GetMapping;

public String hello() {
    return "login";

To process a user login, create another method that handles POST requests with login data. It then redirects the user either to the success or failure page, depending on the result.

Note that the login() method receives a domain object as an argument and returns a ModelAndView object:

import org.springframework.web.bind.annotation.PostMapping;
import org.springframework.web.servlet.ModelAndView;

public ModelAndView login(LoginData loginData) {
    if (LOGIN.equals(loginData.getLogin()) 
      && PASSWORD.equals(loginData.getPassword())) {
        return new ModelAndView("success", 
          Collections.singletonMap("login", loginData.getLogin()));
    } else {
        return new ModelAndView("failure", 
          Collections.singletonMap("login", loginData.getLogin()));

ModelAndView is a holder of two distinct objects:

  • Model – a key-value map of data used to render the page
  • View – a template of the page that is filled with data from the model

These are joined for convenience so that the controller method can return them both at once.

To render your HTML page, use Thymeleaf as a view template engine, which has solid, out-of-the-box integration with Spring.

Servlets as the Foundation of a Java Web Application

So, what does actually happen when you type http://localhost:8080/ in the browser, press Enter, and the request hits the web server? How do you get from this request to seeing a web form in the browser?

Given the project is a simple Spring Boot application, you’ll be able to run it via the Spring5Application.

Spring Boot uses Apache Tomcat by default. Hence, running the application, you are likely to see the following information in the log:

2017-10-16 20:36:11.626  INFO 57414 --- [main] 
  o.s.b.w.embedded.tomcat.TomcatWebServer  : 
  Tomcat initialized with port(s): 8080 (http)

2017-10-16 20:36:11.634  INFO 57414 --- [main] 
  o.apache.catalina.core.StandardService   : 
  Starting service [Tomcat]

2017-10-16 20:36:11.635  INFO 57414 --- [main] 
  org.apache.catalina.core.StandardEngine  : 
  Starting Servlet Engine: Apache Tomcat/8.5.23

Since Tomcat is a Servlet container, naturally every HTTP request sent to a Tomcat web server is processed by a Java servlet. So the Spring Web application entry point is, not surprisingly, a servlet.

A servlet is, simply put, a core component of any Java web application; it’s low-level and does not impose too much in the way of specific programming patterns, such as MVC.

An HTTP servlet can only receive an HTTP request, process it in some way, and send a response back.

And, starting with the Servlet 3.0 API, you can now move beyond XML configuration and start leveraging Java configuration (with minor restrictions).

DispatcherServlet as the Heart of Spring MVC

What we really want to do as developers of a web application is to abstract away the following tedious and boilerplate tasks and focus on useful business logic:

  • mapping an HTTP request to a certain processing method
  • parsing of HTTP request data and headers into data transfer objects (DTOs) or domain objects
  • model-view-controller interaction
  • generation of responses from DTOs, domain objects, etc.

The Spring DispatcherServlet provides exactly that. It is the heart of the Spring Web MVC framework; this core component receives all requests to your application.

As you’ll see, DispatcherServlet is very extensible. For example, it allows you to plug in different existing or new adapters for a lot of tasks:

  • map a request to a class or method that should handle it (implementations of the HandlerMapping interface)
  • handle a request using a specific pattern, like a regular servlet, a more complex MVC workflow, or just a method in a POJO bean (implementations of the HandlerAdapter interface)
  • resolve views by name, allowing you to use different templating engines, XML, XSLT or any other view technology (implementations of the ViewResolver interface)
  • parse multipart requests by using the default Apache Commons file uploading implementation or writing your own MultipartResolver
  • resolve locale with any LocaleResolver implementation, including cookie, session, Accept HTTP header, or any other way of determining the locale expected by the user

Processing of an HTTP Request

First, let’s trace the processing of simple HTTP requests to a method in your controller layer and back to the browser/client.

The DispatcherServlet has a long inheritance hierarchy; it’s worth understanding these individual aspects one by one, top-down. The request processing methods will interest us the most.

Understanding the HTTP request, both locally during standard development, as well as remotely, is a critical part of understanding the MVC architecture.


GenericServlet is a part of the Servlet specification not directly focused on HTTP. It defines the service() method that receives incoming requests and produces responses.

Note how ServletRequest and ServletResponse method arguments are not tied to the HTTP protocol:

public abstract void service(ServletRequest req, ServletResponse res) 
  throws ServletException, IOException;

This is the method that is eventually called on any request to the server, including a simple GET request.


HttpServlet class is, as the name suggests, the HTTP-focused Servlet implementation, also defined by the specification.

In more practical terms, HttpServlet is an abstract class with a service() method implementation that splits the requests by the HTTP method type and looks roughly like this:

protected void service(HttpServletRequest req, HttpServletResponse resp)
    throws ServletException, IOException {

    String method = req.getMethod();
    if (method.equals(METHOD_GET)) {
        // ...
        doGet(req, resp);
    } else if (method.equals(METHOD_HEAD)) {
        // ...
        doHead(req, resp);
    } else if (method.equals(METHOD_POST)) {
        doPost(req, resp);
        // ...


Next, HttpServletBean is the first Spring-aware class in the hierarchy. It injects the bean’s properties using the servlet init-param values received from the web.xml or from WebApplicationInitializer.

In case of the requests to your application, the doGet(), doPost(), etc methods are called for those specific HTTP requests.


FrameworkServlet integrates the Servlet functionality with a web application context, implementing the ApplicationContextAware interface. But it is also able to create a web application context on its own.

As you already saw, the HttpServletBean superclass injects init-params as bean properties. So, if a context class name is provided in the contextClass init-param of the servlet, then an instance of this class will be created as an application context. Otherwise, a default XmlWebApplicationContext class will be used.

As XML configuration is out of style nowadays, Spring Boot configures DispatcherServlet with AnnotationConfigWebApplicationContext by default. But you could change that easily.

For example, if you need to configure your Spring Web MVC application with a Groovy-based application context, you could use the following configuration of DispatcherServlet in the web.xml file:


The same configuration may be done in a more modern Java-based way using the WebApplicationInitializer class.

DispatcherServlet: Unifying the Request Processing

The HttpServlet.service() implementation, which routes requests by the type of HTTP verb, makes perfect sense in the context of low-level servlets. However, at the Spring MVC level of abstraction, method type is just one of the parameters that can be used to map the request to its handler.

And so, the other main function of the FrameworkServlet class is to join the handling logic back into a single processRequest() method, which in turn calls the doService() method:

protected final void doGet(HttpServletRequest request, 
  HttpServletResponse response) throws ServletException, IOException {
    processRequest(request, response);

protected final void doPost(HttpServletRequest request, 
  HttpServletResponse response) throws ServletException, IOException {
    processRequest(request, response);

// …

DispatcherServlet: Enriching the Request

Finally, the DispatcherServlet implements the doService() method. Here, it adds to the request some useful objects that may come in handy down the processing pipeline: web application context, locale resolver, theme resolver, theme source etc.:

request.setAttribute(LOCALE_RESOLVER_ATTRIBUTE, this.localeResolver);
request.setAttribute(THEME_RESOLVER_ATTRIBUTE, this.themeResolver);
request.setAttribute(THEME_SOURCE_ATTRIBUTE, getThemeSource());

Also, the doService() method prepares input and output flash maps. Flash map is basically a pattern to pass parameters from one request to another request that immediately follows. This may be very useful during redirects (like showing the user a one-shot information message after the redirect):

FlashMap inputFlashMap = this.flashMapManager
  .retrieveAndUpdate(request, response);
if (inputFlashMap != null) {
request.setAttribute(OUTPUT_FLASH_MAP_ATTRIBUTE, new FlashMap());

Then, the doService() method calls the doDispatch() method that is responsible for request dispatching.

DispatcherServlet: Dispatching the Request

The main purpose of the dispatch() method is to find an appropriate handler for the request and feed it the request/response parameters. The handler is basically any kind of Object and is not limited to a specific interface. This also means that Spring needs to find an adapter for this handler that knows how to “talk” to the handler.

To find the handler that matches the request, Spring goes through the registered implementations of the HandlerMapping interface. There are many different implementations that could suit your needs.

SimpleUrlHandlerMapping allows mapping a request by its URL to a certain processing bean. For example, it can be configured by injecting its mappings property with a java.util.Properties instance similar to this:


Probably the most widely-used class for handler mapping is RequestMappingHandlerMapping, which maps a request to a @RequestMapping-annotated method of a @Controller class. This is exactly the mapping that connects the dispatcher with the hello() and login() methods of your controller.

Note that your Spring-aware methods are annotated with @GetMapping and @PostMapping correspondingly. These annotations, in turn, are marked with the @RequestMapping meta-annotation.

The dispatch() method also takes care of some other HTTP-specific tasks:

  • short-circuiting processing of the GET request in case the resource was not modified
  • applying the multipart resolver for corresponding requests
  • short-circuiting processing of the request if the handler chose to handle it asynchronously

Handling the Request

Now that Spring determined the handler for the request and the adapter for the handler, it’s time to finally handle the request. Here’s the signature of the HandlerAdapter.handle() method. It’s important to note that the handler has a choice in how to handle the request:

  • write the data to the response object by itself and return null

return a ModelAndView object to be rendered by the DispatcherServlet

ModelAndView handle(HttpServletRequest request, 
                    HttpServletResponse response, 
                    Object handler) throws Exception;

There are several provided types of handlers. Here’s how the SimpleControllerHandlerAdapter processes a Spring MVC controller instance (do not confuse it with a @Controller-annotated POJO).

Notice how the controller handler returns ModelAndView object and does not render the view by itself:

public ModelAndView handle(HttpServletRequest request, 
  HttpServletResponse response, Object handler) throws Exception {
    return ((Controller) handler).handleRequest(request, response);

The second is SimpleServletHandlerAdapter, which adapts a regular Servlet as a request handler.

A Servlet does not know anything about ModelAndView and simply handles the request by itself, rendering the result into the response object. So this adapter simply returns null instead of ModelAndView:

public ModelAndView handle(HttpServletRequest request, 
  HttpServletResponse response, Object handler) throws Exception {
    ((Servlet) handler).service(request, response);
    return null;

In your case, a controller is a POJO with several @RequestMapping annotations, so any handler is basically a method of this class wrapped in a HandlerMethod instance. To adapt to this handler type, Spring uses the RequestMappingHandlerAdapter class.

Processing Arguments and Return Values of Handler Methods

Note that the controller methods do not usually take HttpServletRequest and HttpServletResponse arguments, but instead receive and return many different types of data, such as domain objects, path parameters etc.

Also, note that you are not required to return a ModelAndView instance from a controller method. You may return a view name, or a ResponseEntity or a POJO that will be converted to a JSON response etc.

The RequestMappingHandlerAdapter makes sure the arguments of the method are resolved from the HttpServletRequest. Also, it creates the ModelAndView object from the method’s return value.

There is an important piece of code in the RequestMappingHandlerAdapter that makes sure all this conversion magic takes place:

ServletInvocableHandlerMethod invocableMethod 
  = createInvocableHandlerMethod(handlerMethod);
if (this.argumentResolvers != null) {
if (this.returnValueHandlers != null) {

The argumentResolvers object is a composite of different HandlerMethodArgumentResolver instances.

There are over 30 different argument resolver implementations. They allow extracting of any kind of information from the request and providing it as method arguments. This includes URL path variables, request body parameters, request headers, cookies, session data etc.

The returnValueHandlers object is a composite of HandlerMethodReturnValueHandler objects. There are also a lot of different value handlers that can process the result of your method to create ModelAndView object expected by the adapter.

For instance, when you return a string from the hello() method, the ViewNameMethodReturnValueHandler processes the value. But when you return a ready ModelAndView from the login() method, Spring uses the ModelAndViewMethodReturnValueHandler.

Rendering the View

By now, Spring has processed the HTTP request and received a ModelAndView object, so it has to render the HTML page that the user will see in the browser. It does that based on the model and the selected view encapsulated in the ModelAndView object.

Also note that you could render a JSON object, or XML, or any other data format that can be transferred via HTTP protocol. We’ll touch more on that in the upcoming REST-focused section here.

Let’s get back to the DispatcherServlet. The render() method first sets the response locale using the provided LocaleResolver instance. Let’s assume that your modern browser sets the Accept header correctly and that the AcceptHeaderLocaleResolver is used by default.

During rendering, the ModelAndView object could already contain a reference to a selected view, or just a view name, or nothing at all if the controller was relying on a default view.

Since both hello() and login() methods specify the desired view as a String name, it has to be looked up by this name. So, this is where the viewResolvers list comes into play:

for (ViewResolver viewResolver : this.viewResolvers) {
    View view = viewResolver.resolveViewName(viewName, locale);
    if (view != null) {
        return view;

This is a list of ViewResolver instances, including our ThymeleafViewResolver provided by the thymeleaf-spring5 integration library. This resolver knows where to search for the views, and provides the corresponding view instances.

After calling the view’s render() method, Spring finally completes the request processing by sending the HTML page to the user’s browser:

REST Support

Beyond the typical MVC scenario, we can also use the framework to create REST web services.

Simply put, you can accept a Resource as an input, specify a POJO as a method argument, and annotate it with @RequestBody. You can also annotate the method itself with @ResponseBody to specify that its result has to be transformed directly to an HTTP response:

import org.springframework.web.bind.annotation.RequestBody;
import org.springframework.web.bind.annotation.ResponseBody;

public MyOutputResource sendMessage(
  @RequestBody MyInputResource inputResource) {
    return new MyOutputResource("Received: "
      + inputResource.getRequestMessage());

This is also possible thanks to the extensibility of Spring MVC.

To marshall the internal DTOs to a REST representation, the framework makes use of the HttpMessageConverter infrastructure. For example, one of the implementations is MappingJackson2HttpMessageConverter, which is able to convert model objects to and from JSON using the Jackson library.

And to further simplify the creation of a REST API, Spring introduces the  @RestController annotation. This is handy to assume @ResponseBody semantics by default and avoid explicitly setting that on each REST controller:

import org.springframework.web.bind.annotation.RestController;

public class RestfulWebServiceController {

    public MyOutputResource getMessage() {
        return new MyOutputResource("Hello!");


In this article, you’ve gone through the processing of a request in the Spring MVC framework in detail. You’ve seen how different extensions of the framework work together to provide all the magic and spare you the necessity of handling the tough parts of the HTTP protocol.

Continuously improve your Java applications by using Stackify Retrace, the full lifecycle application performance management tool. Try it out. 

Improve Your Code with Retrace APM

Stackify's APM tools are used by thousands of .NET, Java, PHP, Node.js, Python, & Ruby developers all over the world.
Explore Retrace's product features to learn more.

Learn More

Want to contribute to the Stackify blog?

If you would like to be a guest contributor to the Stackify blog please reach out to [email protected]