Expand my Community achievements bar.

Help shape the future of AI assistance by participating in this quick card sorting activity. Your input will help create a more effective system that better serves your needs and those of your colleagues.

Introduction to the AEP Edge Network

Avatar

Employee

3/12/24

You must have heard of the Edge Network while implementing the Adobe Experience Platform (AEP). However, not everybody knows what it really is or understands it in full. In this post, I want to mitigate this gap and I am going to introduce this part of AEP.

Definition

Let us start with the definition of the concept of "edge network" in general:

It is the physical portion of the network that interconnects with “the outside world.”

I like it because it is concise, and it encapsulates all that needs to be explained. If you think of AEP as a mesh of services, servers, network devices, databases..., most of these elements are hidden behind a security system. This mesh is what we call AEP Core. Only a small portion of this mesh can be accessed from external systems.

aep-architecture-1024.png

The first building block that can be accessed from the outside world is Data Ingestion, but only secured and vetted private networks are allowed. There is also a data access for the Discovery & Insights building block, which requires authentication. The last one is the Edge Network, which is wide open to the world.

Edge Computing

Edge networks have existed for years, with the promise of bringing part of the functionality of a product close to its consumers. The most popular type of edge network is the Content Delivery Network (CDN). In this case, the promise is to keep copies of the content of a web server distributed around the world and having browsers connect to the closest CDN cluster to lower the latency.

Although conceptually similar to a CDN, in practice the AEP Edge Network is quite different from a CDN. AEP is about experiences, which are a combination of audiences and personalized content. Therefore, the AEP Edge Network was created to bring them close to browsers and mobile devices. The goal is that Web SDK and Mobile SDK implementations receive personalized content and send tracking data as fast as possible, reducing the time it takes for a web page or app screen to load.

Capabilities

Now that the concept of edge network is clear, the rest of this post will explain some key features the AEP Edge Network has to offer.

Datastreams

Imagine the lobby of an office building. It has a large, round entry hall. Out of this hall there are multiple corridors, like spokes of a wheel. Each corridor has a unique number. Corridors have multiple doors on both sides, some are open, others are closed. Finally, each door leads to a practice: architects, dentists, accountants…

Now, let us compare this office building with the AEP Edge Network:

  • The lobby would be a node in the AEP Edge Network.
  • Each corridor would be a datastream.
  • The corridor number would be the Datastream ID.
  • The practices would be the AEP Edge Network services.
  • An open door represents an active service.

Following this comparison, a datastream is just a flow of data, connecting sources (browsers, mobile apps…) and destinations, which are called services. The Datastream capability does not modify the input data, it just routes it to the correct service, based on the information contained in the data itself and the Datastream configuration.

At the moment of writing, the following services are available:

  • Adobe Analytics
  • Adobe Audience Manager
  • Adobe Target
  • Event Forwarding
  • Adobe Experience Platform

edge-add-service-1024.png

The first three cases will behave as if you still had AppMeasurement, DIL, or at.js, respectively. The last two are where the AEP Edge Network shines. More on them below.

Profile & Segmentation

This is probably the most important feature of the AEP Edge Network. It is an extension of the Real-Time Customer Profile (RTCP), which resides in the AEP core.

As its name implies, this capability brings profile attributes and segment qualifications to the edge. However, you would be wrong if you thought that you were getting the full RTCP feature on the edge. That would defeat the purpose of an edge network. Instead, when the AEP Edge Network needs a profile, it makes a call to the RTCP and fetches a projection of the profile, which includes profile attributes and segment qualifications.

If this is all this capability did, it would be quite underwhelming. The segmentation would not be real time, but reliant on pre-existing segments in RTCP. The reality is that the AEP Edge Network also includes a real-time segmentation engine. It listens to events coming from a datastream and combines them with profile attributes to qualify for segments on the fly. Not all segment definitions are possible on the edge, refer to the Edge segmentation UI guide for more information.

To enable this capability, you must add the Adobe Experience Platform service to the Datastream you are working with and enable “Edge Segmentation” and “Personalization Destinations.” You must also have one profile Merge Policy enabled for the edge.

edge-aep-service-config-1024.png

Tags

In the architecture diagram at the beginning of this post, you will have noticed that there is another capability on the AEP Edge Network: Tags. This should not confuse you. Tags, formerly known as Launch, are JavaScript libraries executed in the browser, not the AEP Edge Network. However, these libraries are delivered through a CDN. Technically, it is a different network than the rest of the capabilities but, conceptually, they are part of the AEP Edge Network.

In most modern implementations, among the libraries delivered through Tags, you will have the Web SDK to interface with the AEP Edge Network itself.

Event Forwarding

So far, we have been talking about integrations between the AEP Edge Network and Adobe tools. However, Adobe extended the AEP Edge Network functionality to send data to non-Adobe systems. This capability is called Event Forwarding. The goal is that a single call from the browser or app using Web SDK or Mobile SDK should contain all data that needs to be collected by all downstream systems. The AEP Edge Network will forward this data to various destinations, Adobe and non-Adobe, as it can be seen in the diagram below:

event-forwarding-1024.png

It is worth mentioning that Event Forwarding is a paid feature. You will not see it in your Data Collection UI unless you have the corresponding license.

Other capabilities

The features that have been described so far are the most popular and, therefore, have been documented in the architecture diagram at the beginning of this post. However, there are other lesser-known capabilities, and I am sure that Adobe will add more over time.

I will mention two, both based on the Datastream service that connects to the Adobe Experience Platform:

  • Offer Decisioning. In the configuration of the Adobe Experience Platform service, and if you have the corresponding license, there is a checkbox with the name “Offer Decisioning.” This checkbox connects the datastream to the Offer Decisioning Engine (ODE) to get offers. The call to the AEP Edge Network to invoke ODE is similar to the call to invoke Adobe Target.
  • Data Ingestion. It is also possible to stream the events coming from the browser and app into a specific dataset in AEP Core. This data can be used for whatever you want, although the most typical use cases are to generate reports through Customer Journey Analytics, and to trigger and orchestrate journeys in Adobe Journey Optimizer.
6 Comments