Introduction

👍

Beta

Optimizely Edge Agent is in beta. Apply on the Optimizely beta signup page or contact your Customer Success Manager.

Optimizely Edge Agent combines the capabilities of Optimizely's Feature Experimentation edge starter kit SDKs with the versatility and scalability of Optimizely Feature Experimentation Agent.

This hybrid approach lets you seamlessly conduct targeted deliveries and experiments across various platforms and architectures, harnessing the advantages of edge computing and serverless infrastructure.

The two modes of Optimizely Edge Agent

Optimizely Edge Agent lets you optimize your experimentation efforts, reduce latency, and improve the overall performance of your applications by combining the benefits of edge computing and serverless architecture.

Optimizely Edge Agent operates in two distinct modes, depending on the type of request it receives:

• Edge Mode (GET Requests) – Functions as an edge starter SDK when handling GET requests. Makes decisions at the edge before requests are forwarded to the origin or content is retrieved from the origin. It also provides automated handling datafile management automatically, including support for key-value (KV) stores. In this mode, the agent can handle tasks such as:

  • Fetching content from the origin.
  • Caching content.
  • Returning content based on experiment and variation cache keys.
  • Managing cookie serialization.
  • Generating user IDs.
  • Embedding decisions in headers.

• Agent Mode (POST Requests) – Functions as Optimizely Agent when handling POST requests. In this mode, you can interact with an Optimizely Feature Experimentation SDK that functions as a serverless server within your Content Delivery Network (CDN). This lets you obtain decisions remotely without the need to manage infrastructure or worry about scaling. Optimizely Edge Agent handles the heavy lifting, providing a simple and efficient way to access the full functionality of Optimizely Feature Experimentation.

Edge Mode (GET requests)

In Edge Mode, Optimizely Edge Agent offers the following key features:

• Persistent cookie for sticky bucketing – Optimizely Edge Agent can use a persistent cookie like the Feature Experimentation SDK's user profile service. This ensures that users consistently receive the same variation of an experiment across multiple visits, providing a seamless and coherent user experience.

• KV store user profile support – Edge Mode supports a KV store user profile, permitting advanced user targeting and personalization. This lets you store and retrieve user-specific data, enhancing the accuracy and relevance of experimentation decisions.

• Datafile management – Hyrbid Serverless Agent can fetch the datafile from the Optimizely CDN and cache it for improved performance. Or, it can retrieve the datafile from the KV store. Optimizely Edge Agent includes a built-in API to listen to webhooks and automatically update the KV store, ensuring the latest experiment configurations are available.

• Automated decision reconciliation – Optimizely Edge Agent automatically reconciles previous bucketing decisions with the running active experiments included in the datafile. If the serialized cookie decisions or decisions stored in the KV user profile storage cannot be matched to a running experiment, those decisions are removed from the cookie or KV store, and new decisions are included when they are made. This automated process simplifies the management of sticky bucketing and ensures that only relevant and active experiments are considered.

• Flexible configuration – Optimizely Edge Agent is configurable using headers and query parameters for both Edge and Agent modes.

Agent Mode (POST requests)

In Agent Mode, Optimizely Edge Agent offers the following key features:

• Serverless architecture – Agent Mode leverages serverless architecture, eliminating the need for you to manage infrastructure or worry about scaling. It automatically scales to handle varying traffic levels and ensures optimal performance without manual intervention.

• Remote decision making – You can obtain experimentation decisions remotely by interacting with the Feature Experimentation SDK through Optimizely Edge Agent. This lets you access the full functionality of the Optimizely Feature Experimentation platform without the need to embed the SDK directly into your applications.

• Flexible configuration – Optimizely Edge Agent is configurable using headers and query parameters for both Edge and Agent modes. Additionally, it supports JSON body payloads for POST requests in Agent Mode.