SPA and Server-Side Rendering spa-and-server-side-rendering

Overview overview

Single page applications (SPAs) can offer the user a rich, dynamic experience that reacts and behaves in familiar ways, often just like a native application. This is achieved by relying on the client to load the content up front and then do the heavy lifting of handling user interaction and thus minimizing the amount of communication needed between the client and the server, making the app more reactive.

However this can lead to longer initial load times, especially if the SPA is large and rich in its content. To optimize load times, some of the content can be rendered server-side. Using server-side rendering (SSR) can accelerate the initial load of the page and then pass further rendering on to the client.

When to Use SSR when-to-use-ssr

SSR is not required on all projects. Although AEM fully supports JS SSR for SPA, Adobe does not recommend implementing it systematically for every project.

When deciding to implement SSR, you must first estimate what additional complexity, effort, and cost adding SSR realistically represents for the project, including the long-term maintenance. An SSR architecture should be chosen only when the added value clearly exceeds the estimated costs.

SSR usually provides some value when there is a clear “yes” to either of the following questions:

Only when at least one of these two questions is answered with a clear “yes” for your project does Adobe recommend implementing SSR. The following sections describe how to do this using Adobe I/O Runtime.

Adobe I/O Runtime adobe-i-o-runtime

If you are confident that your project requires the implementation of SSR, Adobe’s recommended solution is to use Adobe I/O Runtime.

For more information on Adobe I/O Runtime, see the following:

The following sections detail how Adobe I/O Runtime can be used to implement SSR for your SPA in two different models:

Remote Renderer Configuration remote-renderer-configuration

AEM must know where the remotely-rendered content can be retrieved. Regardless of which model you choose to implement for SSR,, you need to specify to AEM how to access this remote rendering service.

This is done via the RemoteContentRenderer - Configuration Factory OSGi service. Search for the string “RemoteContentRenderer” in the Web Console Configuration console at http://<host>:<port>/system/console/configMgr.

The following fields are available for the configuration:

AEM-Driven Communication Flow aem-driven-communication-flow

When using SSR, the component interaction workflow of SPAs in AEM includes a phase in which the initial content of the app is generated on Adobe I/O Runtime.

Adobe I/O Runtime-Driven Communication Flow adobe-i-o-runtime-driven-communication-flow

The previous section describes the standard and recommended implementation of server-side rendering regarding SPAs in AEM, where AEM performs the bootstrapping and serving of content.

Alternatively, SSR can be implemented so that Adobe I/O Runtime is responsible for the bootstrapping, effectively reversing the communication flow.

Both models are valid and supported by AEM. However, one should consider the advantages and disadvantages of each before implementing a particular model.

Planning for SSR planning-for-ssr

Only part of an application must be rendered server side. The common example is the content that is displayed above the fold on the initial load of the page is rendered server side. This saves time by delivering to the client, already rendered content. As the user interacts with the SPA, the additional content is rendered by the client.

As you consider implementing server-side rendering for your SPA, review for what parts of the app it is necessary.

Developing an SPA using SSR developing-an-spa-using-ssr

SPA components could be rendered by the client (in the browser) or server side. When rendered server side, browser properties such as window size and location are not present. Therefore, SPA components should be isomorphic, making no assumption about where they will be rendered.

To use SSR, deploy your code in AEM and on Adobe I/O Runtime, which is responsible for the server-side rendering. Most of the code will be the same, however, server-specific tasks will differ.

SSR for SPAs in AEM ssr-for-spas-in-aem

SSR for SPAs in AEM require Adobe I/O Runtime, which is called for the rendering of the app content server side. Within the HTL of the app, a resource on Adobe I/O Runtime is called to render the content.

Just as AEM supports the Angular and React SPA frameworks out-of-the box, server-side rendering is also supported for Angular and React apps. See the NPM documentation for both frameworks for further details.

For a simplistic example, see We.Retail Journal app. It renders the entire application server side. Although this is not a real-world example, it does illustrate what is needed to implement SSR.

Using Node.js using-node-js

Adobe I/O Runtime is the recommended solution for implementing SSR for SPAs in AEM.

For on-premesis AEM instances, it is also possible to implement SSR using a custom Node.js instance in the same way as described above. Although this is supported by Adobe, it is not recommended.

Remote Content Renderer remote-content-renderer

The Remote Content Renderer Configuration that is required to use SSR with your SPA in AEM taps into a more generalized rendering service that can be extended and customized to meet your needs.

RemoteContentRenderingService remotecontentrenderingservice

RemoteContentRenderingService is an OSGi service to retrieve content rendered on a remote server, such as from Adobe I/O. The content sent to the remote server is based on the request parameter passed.

RemoteContentRenderingService can be injected by dependency inversion into either a custom Sling model or servlet when additional content manipulation is required.

This service is internally used by the RemoteContentRendererRequestHandlerServlet.

RemoteContentRendererRequestHandlerServlet remotecontentrendererrequesthandlerservlet

The RemoteContentRendererRequestHandlerServlet can be used to programmatically set the request configuration. DefaultRemoteContentRendererRequestHandlerImpl, the provided default request handler implementation, lets you create multiple OSGi configurations to map a location in the content structure to a remote endpoint.

To add a custom request Handler, implement the RemoteContentRendererRequestHandler interface. Be sure to set the Constants.SERVICE_RANKING component property to an integer higher than 100, which is the ranking of the DefaultRemoteContentRendererRequestHandlerImpl.

@Component(immediate = true,
        service = RemoteContentRendererRequestHandler.class,
        property={
            Constants.SERVICE_RANKING +":Integer=1000"
        })
public class CustomRemoteContentRendererRequestHandlerImpl implements RemoteContentRendererRequestHandler {}

Configure the OSGi Configuration of the Default Handler configure-default-handler

The configuration of the default handler must be configured as described in the section Remote Content Renderer Configuration.

Remote Content Renderer Usage usage

To have a servlet fetch and return some content that can be injected into the page:

Usually, the HTL template of a page component is the main recipient of such a feature.

<sly data-sly-resource="${resource @ resourceType='cq/remote/content/renderer/request/handler'}" />

Requirements requirements

The servlets use the Sling Model Exporter to serialize the component data. By default, both the com.adobe.cq.export.json.ContainerExporter and com.adobe.cq.export.json.ComponentExporter are supported as Sling Model adapters. If necessary, you can add classes that the request should be adapted to using the RemoteContentRendererServlet and implementing the RemoteContentRendererRequestHandler#getSlingModelAdapterClasses. The additional classes must extend the ComponentExporter.