21.05

Event-Driven Services: topics

Why Event Driven Architecture

We will discuss and explore the benefits and challenges of Event Driven services and where they fit into modern software architectures. An emphasis will be placed on the particular problems that can be solved by introducing event streaming. Warnings will be highlighted about antipatterns and possible pitfalls and how to avoid them.

Event Driven Microservice Architecture

Getting your head around events and event streaming is a new way of thinking about software architecture for many developers and has repercussions on the architecture of everything from the UI to the database. The good news is many of the benefits of event streaming can be introduced in incremental ways. We will start with a demonstration of a synchronous style service and user interface and show how events can actually simplify the work, particularly at the “edges” and allow application developers to move quickly and focus on delivering value by separating the “what” from “how”. This section will also include a brief introduction to Apache Kafka, and how it is designed as a service, and key concepts and components.

Contracts: Messaging Patterns and Message Design

In this section we will explore mapping the business topology to events, including event identification, definition and design. We will learn the two main types of events, unkeyed events and entity events, and their appropriate usage, including repercussions on scalability. We will discuss, demonstrate, and implement using explicit schema for events, including options such as JSON-Schema, Protobuf, and Apache Avro. We will also look in detail at patterns in Event contract design to make them resilient to versioning, and how to handle contract versioning when the time comes.

Producing and Consuming Events

In this section we will implement a variety of services that either produce events, consume events, or both. We will also discuss DAPR (the Distributed Application Runtime) and how it can insulate service developers from the complexity of managing services that rely on a particular streaming platform. We will explore and explain the guarantees provided for message delivery.

Processing Event Streams

In the Producing and Consuming Events section, our architecture will be a close approximation of a standard Pub/Sub system. The true power of Event Streaming begins to unlock with the introduction of stream processing. We will create both stateful and stateless stream processors, explore standard patterns for processing the stream, and experience how stream processing introduces a form of “reactive programming” that unlocks huge opportunities for delivering value over time. We will create programmatic stream processors, and introduce kSQL as a tool for abstracting stream processing away from imperative code with a special dialect of SQL.

Data Liberation and Materialization

Event streaming allows you to bring the data from the “inside” (traditional, centralized RDBMS) to the “outside” where your applications need it. It also means that new services can be added easily at any time because the obstacles of “walled data” have been removed. The event stream becomes our “source of truth”, and new services can be introduced, and even “time travel” and process events from the past. We will look at ways to connect databases using event streams, including patterns like Change Data Capture.

Integrating Event Driven Services

These next two sections will be an exploration of various techniques for introducing event streaming into existing platforms, integrating with event-ignorant services, and so-called “legacy” systems.

Deploying Event Driven Services

This will be a short topic demonstrating how Event Driven services reinforce the CI/CD “ideal”, and how by moving to Events your organization will be able to have the agility to quickly respond to the needs of the business.